docs: clean up orphaned docs and add session event types

Delete orphaned docs not in docs.json navigation (gigacode.mdx,
foundry-self-hosting.mdx, session-transcript-schema.mdx, pi-support-plan.md).
Remove outdated musl/glibc troubleshooting section. Add event types
documentation with example payloads to agent-sessions.mdx.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.claude/commands/post-release-testing.md

@@ -43,7 +43,7 @@ Manually verify the install script works in a fresh environment:
 ```bash
 docker run --rm alpine:latest sh -c "
   apk add --no-cache curl ca-certificates libstdc++ libgcc bash &&
-  curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh &&
+  curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh &&
   sandbox-agent --version
 "
 ```

.env.development.example

@@ -23,6 +23,9 @@ GITHUB_APP_PRIVATE_KEY=
 # Webhook secret for verifying GitHub webhook payloads.
 # Use smee.io for local development: https://smee.io/new
 GITHUB_WEBHOOK_SECRET=
+# Required for local GitHub webhook forwarding in compose.dev.
+SMEE_URL=
+SMEE_TARGET=http://backend:7741/v1/webhooks/github
 
 # Fill these in when enabling live Stripe billing.
 STRIPE_SECRET_KEY=

.gitignore

@@ -59,3 +59,4 @@ sdks/cli/platforms/*/bin/
 # Foundry desktop app build artifacts
 foundry/packages/desktop/frontend-dist/
 foundry/packages/desktop/src-tauri/sidecars/
+.context/

CLAUDE.md

@@ -1,40 +1,5 @@
 # Instructions
 
-## ACP v1 Baseline
-
-- v1 is ACP-native.
-- `/v1/*` is removed and returns `410 Gone` (`application/problem+json`).
-- `/opencode/*` is disabled during ACP core phases and returns `503`.
-- Prompt/session traffic is ACP JSON-RPC over streamable HTTP on `/v1/rpc`:
-  - `POST /v1/rpc`
-  - `GET /v1/rpc` (SSE)
-  - `DELETE /v1/rpc`
-- Control-plane endpoints:
-  - `GET /v1/health`
-  - `GET /v1/agents`
-  - `POST /v1/agents/{agent}/install`
-- Binary filesystem transfer endpoints (intentionally HTTP, not ACP extension methods):
-  - `GET /v1/fs/file`
-  - `PUT /v1/fs/file`
-  - `POST /v1/fs/upload-batch`
-- Sandbox Agent ACP extension method naming:
-  - Custom ACP methods use `_sandboxagent/...` (not `_sandboxagent/v1/...`).
-  - Session detach method is `_sandboxagent/session/detach`.
-
-## API Scope
-
-- ACP is the primary protocol for agent/session behavior and all functionality that talks directly to the agent.
-- ACP extensions may be used for gaps (for example `skills`, `models`, and related metadata), but the default is that agent-facing behavior is implemented by the agent through ACP.
-- Custom HTTP APIs are for non-agent/session platform services (for example filesystem, terminals, and other host/runtime capabilities).
-- Filesystem and terminal APIs remain Sandbox Agent-specific HTTP contracts and are not ACP.
-  - Do not make Sandbox Agent core flows depend on ACP client implementations of `fs/*` or `terminal/*`; in practice those client-side capabilities are often incomplete or inconsistent.
-  - ACP-native filesystem and terminal methods are also too limited for Sandbox Agent host/runtime needs, so prefer the native HTTP APIs for richer behavior.
-- Keep `GET /v1/fs/file`, `PUT /v1/fs/file`, and `POST /v1/fs/upload-batch` on HTTP:
-  - These are Sandbox Agent host/runtime operations with cross-agent-consistent behavior.
-  - They may involve very large binary transfers that ACP JSON-RPC envelopes are not suited to stream.
-  - This is intentionally separate from ACP native `fs/read_text_file` and `fs/write_text_file`.
-  - ACP extension variants may exist in parallel, but SDK defaults should prefer HTTP for these binary transfer operations.
-
 ## Naming and Ownership
 
 - This repository/product is **Sandbox Agent**.
@@ -48,66 +13,14 @@
 - Never mention "ACP" in user-facing docs (`docs/**/*.mdx`) except in docs that are specifically about ACP itself (e.g. `docs/acp-http-client.mdx`).
 - Never expose underlying protocol method names (e.g. `session/request_permission`, `session/create`, `_sandboxagent/session/detach`) in non-ACP docs. Describe the behavior in user-facing terms instead.
 - Do not describe the underlying protocol implementation in docs. Only document the SDK surface (methods, types, options). ACP protocol details belong exclusively in ACP-specific pages.
+- Do not use em dashes (`—`) in docs. Use commas, periods, or parentheses instead.
 
-## Architecture (Brief)
+### Docs Source Of Truth (HTTP/CLI)
 
-- HTTP contract and problem/error mapping: `server/packages/sandbox-agent/src/router.rs`
-- ACP client runtime and agent process bridge: `server/packages/sandbox-agent/src/acp_runtime/mod.rs`
-- Agent/native + ACP agent process install and lazy install: `server/packages/agent-management/`
-- Inspector UI served at `/ui/` and bound to ACP over HTTP from `frontend/packages/inspector/`
-
-## TypeScript SDK Architecture
-
-- TypeScript clients are split into:
-  - `acp-http-client`: protocol-pure ACP-over-HTTP (`/v1/acp`) with no Sandbox-specific HTTP helpers.
-  - `sandbox-agent`: `SandboxAgent` SDK wrapper that combines ACP session operations with Sandbox control-plane and filesystem helpers.
-- `SandboxAgent` entry points are `SandboxAgent.connect(...)` and `SandboxAgent.start(...)`.
-- Stable Sandbox session methods are `createSession`, `resumeSession`, `resumeOrCreateSession`, `destroySession`, `rawSendSessionMethod`, `onSessionEvent`, `setSessionMode`, `setSessionModel`, `setSessionThoughtLevel`, `setSessionConfigOption`, `getSessionConfigOptions`, `getSessionModes`, `respondPermission`, `rawRespondPermission`, and `onPermissionRequest`.
-- `Session` helpers are `prompt(...)`, `rawSend(...)`, `onEvent(...)`, `setMode(...)`, `setModel(...)`, `setThoughtLevel(...)`, `setConfigOption(...)`, `getConfigOptions()`, `getModes()`, `respondPermission(...)`, `rawRespondPermission(...)`, and `onPermissionRequest(...)`.
-- Cleanup is `sdk.dispose()`.
-
-### React Component Methodology
-
-- Shared React UI belongs in `sdks/react` only when it is reusable outside the Inspector.
-- If the same UI pattern is shared between the Sandbox Agent Inspector and Foundry, prefer extracting it into `sdks/react` instead of maintaining parallel implementations.
-- Keep shared components unstyled by default: behavior in the package, styling in the consumer via `className`, slot-level `classNames`, render overrides, and `data-*` hooks.
-- Prefer extracting reusable pieces such as transcript, composer, and conversation surfaces. Keep Inspector-specific shells such as session selection, session headers, and control-plane actions in `frontend/packages/inspector/`.
-- Document all shared React components in `docs/react-components.mdx`, and keep that page aligned with the exported surface in `sdks/react/src/index.ts`.
-
-### TypeScript SDK Naming Conventions
-
-- Use `respond<Thing>(id, reply)` for SDK methods that reply to an agent-initiated request (e.g. `respondPermission`). This is the standard pattern for answering any inbound JSON-RPC request from the agent.
-- Prefix raw/low-level escape hatches with `raw` (e.g. `rawRespondPermission`, `rawSend`). These accept protocol-level types directly and bypass SDK abstractions.
-
-### Docs Source Of Truth
-
-- For TypeScript docs/examples, source of truth is implementation in:
-  - `sdks/typescript/src/client.ts`
-  - `sdks/typescript/src/index.ts`
-  - `sdks/acp-http-client/src/index.ts`
-- Do not document TypeScript APIs unless they are exported and implemented in those files.
 - For HTTP/CLI docs/examples, source of truth is:
   - `server/packages/sandbox-agent/src/router.rs`
   - `server/packages/sandbox-agent/src/cli.rs`
-- Keep docs aligned to implemented endpoints/commands only (for example ACP under `/v1/acp`, not legacy `/v1/sessions` APIs).
+- Keep docs aligned to implemented endpoints/commands only (for example ACP under `/v1/acp`, not legacy session REST APIs).
-
-## ACP Protocol Compliance
-
-- Before adding any new ACP method, property, or config option category to the SDK, verify it exists in the ACP spec at `https://agentclientprotocol.com/llms-full.txt`.
-- Valid `SessionConfigOptionCategory` values are: `mode`, `model`, `thought_level`, `other`, or custom categories prefixed with `_` (e.g. `_permission_mode`).
-- Do not invent ACP properties or categories (e.g. `permission_mode` is not a valid ACP category — use `_permission_mode` if it's a custom extension, or use existing ACP mechanisms like `session/set_mode`).
-- `NewSessionRequest` only has `_meta`, `cwd`, and `mcpServers`. Do not add non-ACP fields to it.
-- Sandbox Agent SDK abstractions (like `SessionCreateRequest`) may add convenience properties, but must clearly map to real ACP methods internally and not send fabricated fields over the wire.
-
-## Source Documents
-
-- ACP protocol specification (full LLM-readable reference): `https://agentclientprotocol.com/llms-full.txt`
-- `~/misc/acp-docs/schema/schema.json`
-- `~/misc/acp-docs/schema/meta.json`
-- `research/acp/spec.md`
-- `research/acp/v1-schema-to-acp-mapping.md`
-- `research/acp/friction.md`
-- `research/acp/todo.md`
 
 ## Change Tracking
 
@@ -119,14 +32,20 @@
 - Append blockers/decisions to `research/acp/friction.md` during ACP work.
 - `docs/agent-capabilities.mdx` lists models/modes/thought levels per agent. Update it when adding a new agent or changing `fallback_config_options`. If its "Last updated" date is >2 weeks old, re-run `cd scripts/agent-configs && npx tsx dump.ts` and update the doc to match. Source data: `scripts/agent-configs/resources/*.json` and hardcoded entries in `server/packages/sandbox-agent/src/router/support.rs` (`fallback_config_options`).
 - Some agent models are gated by subscription (e.g. Claude `opus`). The live report only shows models available to the current credentials. The static doc and JSON resource files should list all known models regardless of subscription tier.
-- TypeScript SDK tests should run against a real running server/runtime over real `/v1` HTTP APIs, typically using the real `mock` agent for deterministic behavior.
-- Do not use Vitest fetch/transport mocks to simulate server functionality in TypeScript SDK tests.
 
-## Docker Examples (Dev Testing)
+## Docker Test Image
 
-- When manually testing bleeding-edge (unreleased) versions of sandbox-agent in `examples/`, use `SANDBOX_AGENT_DEV=1` with the Docker-based examples.
+- Docker-backed Rust and TypeScript tests build `docker/test-agent/Dockerfile` directly in-process and cache the image tag only in memory (`OnceLock` in Rust, module-level variable in TypeScript).
-- This triggers a local build of `docker/runtime/Dockerfile.full` which builds the server binary from local source and packages it into the Docker image.
+- Do not add cross-process image-build scripts unless there is a concrete need for them.
-- Example: `SANDBOX_AGENT_DEV=1 pnpm --filter @sandbox-agent/example-mcp start`
+
+## Common Software Sync
+
+- These three files must stay in sync:
+  - `docs/common-software.mdx` (user-facing documentation)
+  - `docker/test-common-software/Dockerfile` (packages installed in the test image)
+  - `server/packages/sandbox-agent/tests/common_software.rs` (test assertions)
+- When adding or removing software from `docs/common-software.mdx`, also add/remove the corresponding `apt-get install` line in the Dockerfile and add/remove the test in `common_software.rs`.
+- Run `cargo test -p sandbox-agent --test common_software` to verify.
 
 ## Install Version References
 
@@ -159,4 +78,3 @@
   - `scripts/release/main.ts`
   - `scripts/release/promote-artifacts.ts`
   - `scripts/release/sdk.ts`
-  - `scripts/sandbox-testing/test-sandbox.ts`

Cargo.toml

@@ -4,7 +4,7 @@ members = ["server/packages/*", "gigacode"]
 exclude = ["factory/packages/desktop/src-tauri", "foundry/packages/desktop/src-tauri"]
 
 [workspace.package]
-version = "0.3.2"
+version = "0.4.2"
 edition = "2021"
 authors = [ "Rivet Gaming, LLC <developer@rivet.gg>" ]
 license = "Apache-2.0"
@@ -13,13 +13,13 @@ description = "Universal API for automatic coding agents in sandboxes. Supports
 
 [workspace.dependencies]
 # Internal crates
-sandbox-agent = { version = "0.3.2", path = "server/packages/sandbox-agent" }
+sandbox-agent = { version = "0.4.2", path = "server/packages/sandbox-agent" }
-sandbox-agent-error = { version = "0.3.2", path = "server/packages/error" }
+sandbox-agent-error = { version = "0.4.2", path = "server/packages/error" }
-sandbox-agent-agent-management = { version = "0.3.2", path = "server/packages/agent-management" }
+sandbox-agent-agent-management = { version = "0.4.2", path = "server/packages/agent-management" }
-sandbox-agent-agent-credentials = { version = "0.3.2", path = "server/packages/agent-credentials" }
+sandbox-agent-agent-credentials = { version = "0.4.2", path = "server/packages/agent-credentials" }
-sandbox-agent-opencode-adapter = { version = "0.3.2", path = "server/packages/opencode-adapter" }
+sandbox-agent-opencode-adapter = { version = "0.4.2", path = "server/packages/opencode-adapter" }
-sandbox-agent-opencode-server-manager = { version = "0.3.2", path = "server/packages/opencode-server-manager" }
+sandbox-agent-opencode-server-manager = { version = "0.4.2", path = "server/packages/opencode-server-manager" }
-acp-http-adapter = { version = "0.3.2", path = "server/packages/acp-http-adapter" }
+acp-http-adapter = { version = "0.4.2", path = "server/packages/acp-http-adapter" }
 
 # Serialization
 serde = { version = "1.0", features = ["derive"] }

README.md

@@ -80,11 +80,11 @@ Import the SDK directly into your Node or browser application. Full type safety
 **Install**
 
 ```bash
-npm install sandbox-agent@0.3.x
+npm install sandbox-agent@0.4.x
 ```
 
 ```bash
-bun add sandbox-agent@0.3.x
+bun add sandbox-agent@0.4.x
 # Optional: allow Bun to run postinstall scripts for native binaries (required for SandboxAgent.start()).
 bun pm trust @sandbox-agent/cli-linux-x64 @sandbox-agent/cli-linux-arm64 @sandbox-agent/cli-darwin-arm64 @sandbox-agent/cli-darwin-x64 @sandbox-agent/cli-win32-x64
 ```
@@ -135,7 +135,7 @@ Run as an HTTP server and connect from any language. Deploy to E2B, Daytona, Ver
 
 ```bash
 # Install it
-curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh
+curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh
 # Run it
 sandbox-agent server --token "$SANDBOX_TOKEN" --host 127.0.0.1 --port 2468
 ```
@@ -159,12 +159,12 @@ sandbox-agent server --no-token --host 127.0.0.1 --port 2468
 Install the CLI wrapper (optional but convenient):
 
 ```bash
-npm install -g @sandbox-agent/cli@0.3.x
+npm install -g @sandbox-agent/cli@0.4.x
 ```
 
 ```bash
 # Allow Bun to run postinstall scripts for native binaries.
-bun add -g @sandbox-agent/cli@0.3.x
+bun add -g @sandbox-agent/cli@0.4.x
 bun pm -g trust @sandbox-agent/cli-linux-x64 @sandbox-agent/cli-linux-arm64 @sandbox-agent/cli-darwin-arm64 @sandbox-agent/cli-darwin-x64 @sandbox-agent/cli-win32-x64
 ```
 
@@ -179,11 +179,11 @@ sandbox-agent api sessions send-message-stream my-session --message "Hello" --en
 You can also use npx like:
 
 ```bash
-npx @sandbox-agent/cli@0.3.x --help
+npx @sandbox-agent/cli@0.4.x --help
 ```
 
 ```bash
-bunx @sandbox-agent/cli@0.3.x --help
+bunx @sandbox-agent/cli@0.4.x --help
 ```
 
 [CLI documentation](https://sandboxagent.dev/docs/cli)
@@ -277,7 +277,7 @@ Coding agents expect interactive terminals with proper TTY handling. SSH with pi
 - **Storage of sessions on disk**: Sessions are already stored by the respective coding agents on disk. It's assumed that the consumer is streaming data from this machine to an external storage, such as Postgres, ClickHouse, or Rivet.
 - **Direct LLM wrappers**: Use the [Vercel AI SDK](https://ai-sdk.dev/docs/introduction) if you want to implement your own agent from scratch.
 - **Git Repo Management**: Just use git commands or the features provided by your sandbox provider of choice.
-- **Sandbox Provider API**: Sandbox providers have many nuanced differences in their API, it does not make sense for us to try to provide a custom layer. Instead, we opt to provide guides that let you integrate this project with sandbox providers.
+- **Sandbox Provider API**: Sandbox providers have many nuanced differences in their API, it does not make sense for us to try to provide a custom layer. Instead, we opt to provide guides that let you integrate this repository with sandbox providers.
 
 ## Roadmap
 

docker/inspector-dev/Dockerfile

@@ -0,0 +1,7 @@
+FROM node:22-bookworm-slim
+
+RUN npm install -g pnpm@10.28.2
+
+WORKDIR /app
+
+CMD ["bash", "-lc", "pnpm install --filter @sandbox-agent/inspector... && cd frontend/packages/inspector && exec pnpm vite --host 0.0.0.0 --port 5173"]

docker/release/linux-aarch64.Dockerfile

@@ -10,7 +10,6 @@ COPY package.json pnpm-lock.yaml pnpm-workspace.yaml ./
 COPY frontend/packages/inspector/package.json ./frontend/packages/inspector/
 COPY sdks/cli-shared/package.json ./sdks/cli-shared/
 COPY sdks/acp-http-client/package.json ./sdks/acp-http-client/
-COPY sdks/persist-indexeddb/package.json ./sdks/persist-indexeddb/
 COPY sdks/react/package.json ./sdks/react/
 COPY sdks/typescript/package.json ./sdks/typescript/
 
@@ -21,15 +20,13 @@ RUN pnpm install --filter @sandbox-agent/inspector...
 COPY docs/openapi.json ./docs/
 COPY sdks/cli-shared ./sdks/cli-shared
 COPY sdks/acp-http-client ./sdks/acp-http-client
-COPY sdks/persist-indexeddb ./sdks/persist-indexeddb
 COPY sdks/react ./sdks/react
 COPY sdks/typescript ./sdks/typescript
 
-# Build cli-shared, acp-http-client, SDK, then persist-indexeddb and react (depends on SDK)
+# Build cli-shared, acp-http-client, SDK, then react (depends on SDK)
 RUN cd sdks/cli-shared && pnpm exec tsup
 RUN cd sdks/acp-http-client && pnpm exec tsup
 RUN cd sdks/typescript && SKIP_OPENAPI_GEN=1 pnpm exec tsup
-RUN cd sdks/persist-indexeddb && pnpm exec tsup
 RUN cd sdks/react && pnpm exec tsup
 
 # Copy inspector source and build

docker/release/linux-x86_64.Dockerfile

@@ -10,7 +10,6 @@ COPY package.json pnpm-lock.yaml pnpm-workspace.yaml ./
 COPY frontend/packages/inspector/package.json ./frontend/packages/inspector/
 COPY sdks/cli-shared/package.json ./sdks/cli-shared/
 COPY sdks/acp-http-client/package.json ./sdks/acp-http-client/
-COPY sdks/persist-indexeddb/package.json ./sdks/persist-indexeddb/
 COPY sdks/react/package.json ./sdks/react/
 COPY sdks/typescript/package.json ./sdks/typescript/
 
@@ -21,15 +20,13 @@ RUN pnpm install --filter @sandbox-agent/inspector...
 COPY docs/openapi.json ./docs/
 COPY sdks/cli-shared ./sdks/cli-shared
 COPY sdks/acp-http-client ./sdks/acp-http-client
-COPY sdks/persist-indexeddb ./sdks/persist-indexeddb
 COPY sdks/react ./sdks/react
 COPY sdks/typescript ./sdks/typescript
 
-# Build cli-shared, acp-http-client, SDK, then persist-indexeddb and react (depends on SDK)
+# Build cli-shared, acp-http-client, SDK, then react (depends on SDK)
 RUN cd sdks/cli-shared && pnpm exec tsup
 RUN cd sdks/acp-http-client && pnpm exec tsup
 RUN cd sdks/typescript && SKIP_OPENAPI_GEN=1 pnpm exec tsup
-RUN cd sdks/persist-indexeddb && pnpm exec tsup
 RUN cd sdks/react && pnpm exec tsup
 
 # Copy inspector source and build

docker/release/macos-aarch64.Dockerfile

@@ -10,7 +10,6 @@ COPY package.json pnpm-lock.yaml pnpm-workspace.yaml ./
 COPY frontend/packages/inspector/package.json ./frontend/packages/inspector/
 COPY sdks/cli-shared/package.json ./sdks/cli-shared/
 COPY sdks/acp-http-client/package.json ./sdks/acp-http-client/
-COPY sdks/persist-indexeddb/package.json ./sdks/persist-indexeddb/
 COPY sdks/react/package.json ./sdks/react/
 COPY sdks/typescript/package.json ./sdks/typescript/
 
@@ -21,15 +20,13 @@ RUN pnpm install --filter @sandbox-agent/inspector...
 COPY docs/openapi.json ./docs/
 COPY sdks/cli-shared ./sdks/cli-shared
 COPY sdks/acp-http-client ./sdks/acp-http-client
-COPY sdks/persist-indexeddb ./sdks/persist-indexeddb
 COPY sdks/react ./sdks/react
 COPY sdks/typescript ./sdks/typescript
 
-# Build cli-shared, acp-http-client, SDK, then persist-indexeddb and react (depends on SDK)
+# Build cli-shared, acp-http-client, SDK, then react (depends on SDK)
 RUN cd sdks/cli-shared && pnpm exec tsup
 RUN cd sdks/acp-http-client && pnpm exec tsup
 RUN cd sdks/typescript && SKIP_OPENAPI_GEN=1 pnpm exec tsup
-RUN cd sdks/persist-indexeddb && pnpm exec tsup
 RUN cd sdks/react && pnpm exec tsup
 
 # Copy inspector source and build

docker/release/macos-x86_64.Dockerfile

@@ -10,7 +10,6 @@ COPY package.json pnpm-lock.yaml pnpm-workspace.yaml ./
 COPY frontend/packages/inspector/package.json ./frontend/packages/inspector/
 COPY sdks/cli-shared/package.json ./sdks/cli-shared/
 COPY sdks/acp-http-client/package.json ./sdks/acp-http-client/
-COPY sdks/persist-indexeddb/package.json ./sdks/persist-indexeddb/
 COPY sdks/react/package.json ./sdks/react/
 COPY sdks/typescript/package.json ./sdks/typescript/
 
@@ -21,15 +20,13 @@ RUN pnpm install --filter @sandbox-agent/inspector...
 COPY docs/openapi.json ./docs/
 COPY sdks/cli-shared ./sdks/cli-shared
 COPY sdks/acp-http-client ./sdks/acp-http-client
-COPY sdks/persist-indexeddb ./sdks/persist-indexeddb
 COPY sdks/react ./sdks/react
 COPY sdks/typescript ./sdks/typescript
 
-# Build cli-shared, acp-http-client, SDK, then persist-indexeddb and react (depends on SDK)
+# Build cli-shared, acp-http-client, SDK, then react (depends on SDK)
 RUN cd sdks/cli-shared && pnpm exec tsup
 RUN cd sdks/acp-http-client && pnpm exec tsup
 RUN cd sdks/typescript && SKIP_OPENAPI_GEN=1 pnpm exec tsup
-RUN cd sdks/persist-indexeddb && pnpm exec tsup
 RUN cd sdks/react && pnpm exec tsup
 
 # Copy inspector source and build

docker/release/windows.Dockerfile

@@ -10,7 +10,6 @@ COPY package.json pnpm-lock.yaml pnpm-workspace.yaml ./
 COPY frontend/packages/inspector/package.json ./frontend/packages/inspector/
 COPY sdks/cli-shared/package.json ./sdks/cli-shared/
 COPY sdks/acp-http-client/package.json ./sdks/acp-http-client/
-COPY sdks/persist-indexeddb/package.json ./sdks/persist-indexeddb/
 COPY sdks/react/package.json ./sdks/react/
 COPY sdks/typescript/package.json ./sdks/typescript/
 
@@ -21,15 +20,13 @@ RUN pnpm install --filter @sandbox-agent/inspector...
 COPY docs/openapi.json ./docs/
 COPY sdks/cli-shared ./sdks/cli-shared
 COPY sdks/acp-http-client ./sdks/acp-http-client
-COPY sdks/persist-indexeddb ./sdks/persist-indexeddb
 COPY sdks/react ./sdks/react
 COPY sdks/typescript ./sdks/typescript
 
-# Build cli-shared, acp-http-client, SDK, then persist-indexeddb and react (depends on SDK)
+# Build cli-shared, acp-http-client, SDK, then react (depends on SDK)
 RUN cd sdks/cli-shared && pnpm exec tsup
 RUN cd sdks/acp-http-client && pnpm exec tsup
 RUN cd sdks/typescript && SKIP_OPENAPI_GEN=1 pnpm exec tsup
-RUN cd sdks/persist-indexeddb && pnpm exec tsup
 RUN cd sdks/react && pnpm exec tsup
 
 # Copy inspector source and build

docker/runtime/Dockerfile

@@ -12,7 +12,6 @@ COPY package.json pnpm-lock.yaml pnpm-workspace.yaml ./
 COPY frontend/packages/inspector/package.json ./frontend/packages/inspector/
 COPY sdks/cli-shared/package.json ./sdks/cli-shared/
 COPY sdks/acp-http-client/package.json ./sdks/acp-http-client/
-COPY sdks/persist-indexeddb/package.json ./sdks/persist-indexeddb/
 COPY sdks/react/package.json ./sdks/react/
 COPY sdks/typescript/package.json ./sdks/typescript/
 
@@ -23,7 +22,6 @@ RUN pnpm install --filter @sandbox-agent/inspector...
 COPY docs/openapi.json ./docs/
 COPY sdks/cli-shared ./sdks/cli-shared
 COPY sdks/acp-http-client ./sdks/acp-http-client
-COPY sdks/persist-indexeddb ./sdks/persist-indexeddb
 COPY sdks/react ./sdks/react
 COPY sdks/typescript ./sdks/typescript
 
@@ -31,7 +29,6 @@ COPY sdks/typescript ./sdks/typescript
 RUN cd sdks/cli-shared && pnpm exec tsup
 RUN cd sdks/acp-http-client && pnpm exec tsup
 RUN cd sdks/typescript && SKIP_OPENAPI_GEN=1 pnpm exec tsup
-RUN cd sdks/persist-indexeddb && pnpm exec tsup
 RUN cd sdks/react && pnpm exec tsup
 
 # Copy inspector source and build
@@ -152,7 +149,8 @@ FROM debian:bookworm-slim
 RUN apt-get update && apt-get install -y \
     ca-certificates \
     curl \
-    git && \
+    git \
+    ffmpeg && \
     rm -rf /var/lib/apt/lists/*
 
 # Copy the binary from builder

docker/runtime/Dockerfile.full

@@ -11,7 +11,6 @@ COPY package.json pnpm-lock.yaml pnpm-workspace.yaml ./
 COPY frontend/packages/inspector/package.json ./frontend/packages/inspector/
 COPY sdks/cli-shared/package.json ./sdks/cli-shared/
 COPY sdks/acp-http-client/package.json ./sdks/acp-http-client/
-COPY sdks/persist-indexeddb/package.json ./sdks/persist-indexeddb/
 COPY sdks/react/package.json ./sdks/react/
 COPY sdks/typescript/package.json ./sdks/typescript/
 
@@ -20,14 +19,12 @@ RUN pnpm install --filter @sandbox-agent/inspector...
 COPY docs/openapi.json ./docs/
 COPY sdks/cli-shared ./sdks/cli-shared
 COPY sdks/acp-http-client ./sdks/acp-http-client
-COPY sdks/persist-indexeddb ./sdks/persist-indexeddb
 COPY sdks/react ./sdks/react
 COPY sdks/typescript ./sdks/typescript
 
 RUN cd sdks/cli-shared && pnpm exec tsup
 RUN cd sdks/acp-http-client && pnpm exec tsup
 RUN cd sdks/typescript && SKIP_OPENAPI_GEN=1 pnpm exec tsup
-RUN cd sdks/persist-indexeddb && pnpm exec tsup
 RUN cd sdks/react && pnpm exec tsup
 
 COPY frontend/packages/inspector ./frontend/packages/inspector

docker/test-agent/Dockerfile

@@ -0,0 +1,61 @@
+FROM rust:1.88.0-bookworm AS builder
+WORKDIR /build
+
+COPY Cargo.toml Cargo.lock ./
+COPY server/ ./server/
+COPY gigacode/ ./gigacode/
+COPY resources/agent-schemas/artifacts/ ./resources/agent-schemas/artifacts/
+COPY scripts/agent-configs/ ./scripts/agent-configs/
+COPY scripts/audit-acp-deps/ ./scripts/audit-acp-deps/
+
+ENV SANDBOX_AGENT_SKIP_INSPECTOR=1
+
+RUN --mount=type=cache,target=/usr/local/cargo/registry \
+    --mount=type=cache,target=/usr/local/cargo/git \
+    --mount=type=cache,target=/build/target \
+    cargo build -p sandbox-agent --release && \
+    cp target/release/sandbox-agent /sandbox-agent
+
+# Extract neko binary from the official image for WebRTC desktop streaming.
+# Using neko v3 base image from GHCR which provides multi-arch support (amd64, arm64).
+# Pinned by digest to prevent breaking changes from upstream.
+# Reference client: https://github.com/demodesk/neko-client/blob/37f93eae6bd55b333c94bd009d7f2b079075a026/src/component/internal/webrtc.ts
+FROM ghcr.io/m1k1o/neko/base@sha256:0c384afa56268aaa2d5570211d284763d0840dcdd1a7d9a24be3081d94d3dfce AS neko-base
+
+FROM node:22-bookworm-slim
+RUN apt-get update -qq && \
+    apt-get install -y -qq --no-install-recommends \
+      ca-certificates \
+      bash \
+      libstdc++6 \
+      xvfb \
+      openbox \
+      xdotool \
+      imagemagick \
+      ffmpeg \
+      gstreamer1.0-tools \
+      gstreamer1.0-plugins-base \
+      gstreamer1.0-plugins-good \
+      gstreamer1.0-plugins-bad \
+      gstreamer1.0-plugins-ugly \
+      gstreamer1.0-nice \
+      gstreamer1.0-x \
+      gstreamer1.0-pulseaudio \
+      libxcvt0 \
+      x11-xserver-utils \
+      dbus-x11 \
+      xauth \
+      fonts-dejavu-core \
+      xterm \
+      > /dev/null 2>&1 && \
+    rm -rf /var/lib/apt/lists/*
+
+COPY --from=builder /sandbox-agent /usr/local/bin/sandbox-agent
+COPY --from=neko-base /usr/bin/neko /usr/local/bin/neko
+
+EXPOSE 3000
+# Expose UDP port range for WebRTC media transport
+EXPOSE 59050-59070/udp
+
+ENTRYPOINT ["/usr/local/bin/sandbox-agent"]
+CMD ["server", "--host", "0.0.0.0", "--port", "3000", "--no-token"]

docker/test-common-software/Dockerfile

@@ -0,0 +1,37 @@
+# Extends the base test-agent image with common software pre-installed.
+# Used by the common_software integration test to verify that all documented
+# software in docs/common-software.mdx works correctly inside the sandbox.
+#
+# KEEP IN SYNC with docs/common-software.mdx
+
+ARG BASE_IMAGE=sandbox-agent-test:dev
+FROM ${BASE_IMAGE}
+
+USER root
+
+RUN apt-get update -qq && \
+    apt-get install -y -qq --no-install-recommends \
+      # Browsers
+      chromium \
+      firefox-esr \
+      # Languages
+      python3 python3-pip python3-venv \
+      default-jdk \
+      ruby-full \
+      # Databases
+      sqlite3 \
+      redis-server \
+      # Build tools
+      build-essential cmake pkg-config \
+      # CLI tools
+      git jq tmux \
+      # Media and graphics
+      imagemagick \
+      poppler-utils \
+      # Desktop apps
+      gimp \
+      > /dev/null 2>&1 && \
+    rm -rf /var/lib/apt/lists/*
+
+ENTRYPOINT ["/usr/local/bin/sandbox-agent"]
+CMD ["server", "--host", "0.0.0.0", "--port", "3000", "--no-token"]

docs/agent-capabilities.mdx

@@ -1,127 +0,0 @@
----
-title: "Agent Capabilities"
-description: "Models, modes, and thought levels supported by each agent."
----
-
-Capabilities are subject to change as the agents are updated. See [Agent Sessions](/agent-sessions) for full session configuration API details.
-
-
-<Info>
-    _Last updated: March 5th, 2026. See [Generating a live report](#generating-a-live-report) for up-to-date reference._
-</Info>
-
-## Claude
-
-| Category | Values |
-|----------|--------|
-| **Models** | `default`, `sonnet`, `opus`, `haiku` |
-| **Modes** | `default`, `acceptEdits`, `plan`, `dontAsk`, `bypassPermissions` |
-| **Thought levels** | Unsupported |
-
-### Configuring Effort Level For Claude
-
-Claude does not natively support changing effort level after a session starts, so configure it in the filesystem before creating the session.
-
-```ts
-import { mkdir, writeFile } from "node:fs/promises";
-import path from "node:path";
-import { SandboxAgent } from "sandbox-agent";
-
-const cwd = "/path/to/workspace";
-await mkdir(path.join(cwd, ".claude"), { recursive: true });
-await writeFile(
-  path.join(cwd, ".claude", "settings.json"),
-  JSON.stringify({ effortLevel: "high" }, null, 2),
-);
-
-const sdk = await SandboxAgent.connect({ baseUrl: "http://127.0.0.1:2468" });
-await sdk.createSession({
-  agent: "claude",
-  sessionInit: { cwd, mcpServers: [] },
-});
-```
-
-<Accordion title="Supported file locations (highest precedence last)">
-
-1. `~/.claude/settings.json`
-2. `<session cwd>/.claude/settings.json`
-3. `<session cwd>/.claude/settings.local.json`
-
-</Accordion>
-
-## Codex
-
-| Category | Values |
-|----------|--------|
-| **Models** | `gpt-5.3-codex` (default), `gpt-5.3-codex-spark`, `gpt-5.2-codex`, `gpt-5.1-codex-max`, `gpt-5.2`, `gpt-5.1-codex-mini` |
-| **Modes** | `read-only` (default), `auto`, `full-access` |
-| **Thought levels** | `low`, `medium`, `high` (default), `xhigh` |
-
-## OpenCode
-
-| Category | Values |
-|----------|--------|
-| **Models** | See below |
-| **Modes** | `build` (default), `plan` |
-| **Thought levels** | Unsupported |
-
-<Accordion title="See all models">
-
-| Provider | Models |
-|----------|--------|
-| **Anthropic** | `anthropic/claude-3-5-haiku-20241022`, `anthropic/claude-3-5-haiku-latest`, `anthropic/claude-3-5-sonnet-20240620`, `anthropic/claude-3-5-sonnet-20241022`, `anthropic/claude-3-7-sonnet-20250219`, `anthropic/claude-3-7-sonnet-latest`, `anthropic/claude-3-haiku-20240307`, `anthropic/claude-3-opus-20240229`, `anthropic/claude-3-sonnet-20240229`, `anthropic/claude-haiku-4-5`, `anthropic/claude-haiku-4-5-20251001`, `anthropic/claude-opus-4-0`, `anthropic/claude-opus-4-1`, `anthropic/claude-opus-4-1-20250805`, `anthropic/claude-opus-4-20250514`, `anthropic/claude-opus-4-5`, `anthropic/claude-opus-4-5-20251101`, `anthropic/claude-opus-4-6`, `anthropic/claude-sonnet-4-0`, `anthropic/claude-sonnet-4-20250514`, `anthropic/claude-sonnet-4-5`, `anthropic/claude-sonnet-4-5-20250929` |
-| **OpenAI** | `openai/gpt-5.1-codex`, `openai/gpt-5.1-codex-max`, `openai/gpt-5.1-codex-mini`, `openai/gpt-5.2`, `openai/gpt-5.2-codex`, `openai/gpt-5.3-codex` |
-| **Cerebras** | `cerebras/gpt-oss-120b`, `cerebras/qwen-3-235b-a22b-instruct-2507`, `cerebras/zai-glm-4.7` |
-| **OpenCode Zen** | `opencode/big-pickle`, `opencode/claude-3-5-haiku`, `opencode/claude-haiku-4-5`, `opencode/claude-opus-4-1`, `opencode/claude-opus-4-5`, `opencode/claude-opus-4-6`, `opencode/claude-sonnet-4`, `opencode/claude-sonnet-4-5`, `opencode/gemini-3-flash`, `opencode/gemini-3-pro` (default), `opencode/glm-4.6`, `opencode/glm-4.7`, `opencode/gpt-5`, `opencode/gpt-5-codex`, `opencode/gpt-5-nano`, `opencode/gpt-5.1`, `opencode/gpt-5.1-codex`, `opencode/gpt-5.1-codex-max`, `opencode/gpt-5.1-codex-mini`, `opencode/gpt-5.2`, `opencode/gpt-5.2-codex`, `opencode/kimi-k2`, `opencode/kimi-k2-thinking`, `opencode/kimi-k2.5`, `opencode/kimi-k2.5-free`, `opencode/minimax-m2.1`, `opencode/minimax-m2.1-free`, `opencode/trinity-large-preview-free` |
-
-</Accordion>
-
-## Cursor
-
-| Category | Values |
-|----------|--------|
-| **Models** | See below |
-| **Modes** | Unsupported |
-| **Thought levels** | Unsupported |
-
-<Accordion title="See all models">
-
-| Group | Models |
-|-------|--------|
-| **Auto** | `auto` |
-| **Composer** | `composer-1.5`, `composer-1` |
-| **GPT-5.3 Codex** | `gpt-5.3-codex`, `gpt-5.3-codex-low`, `gpt-5.3-codex-high`, `gpt-5.3-codex-xhigh`, `gpt-5.3-codex-fast`, `gpt-5.3-codex-low-fast`, `gpt-5.3-codex-high-fast`, `gpt-5.3-codex-xhigh-fast` |
-| **GPT-5.2** | `gpt-5.2`, `gpt-5.2-high`, `gpt-5.2-codex`, `gpt-5.2-codex-low`, `gpt-5.2-codex-high`, `gpt-5.2-codex-xhigh`, `gpt-5.2-codex-fast`, `gpt-5.2-codex-low-fast`, `gpt-5.2-codex-high-fast`, `gpt-5.2-codex-xhigh-fast` |
-| **GPT-5.1** | `gpt-5.1-high`, `gpt-5.1-codex-max`, `gpt-5.1-codex-max-high` |
-| **Claude** | `opus-4.6-thinking` (default), `opus-4.6`, `opus-4.5`, `opus-4.5-thinking`, `sonnet-4.5`, `sonnet-4.5-thinking` |
-| **Other** | `gemini-3-pro`, `gemini-3-flash`, `grok` |
-
-</Accordion>
-
-## Amp
-
-| Category | Values |
-|----------|--------|
-| **Models** | `amp-default` |
-| **Modes** | `default`, `bypass` |
-| **Thought levels** | Unsupported |
-
-## Pi
-
-| Category | Values |
-|----------|--------|
-| **Models** | `default` |
-| **Modes** | Unsupported |
-| **Thought levels** | Unsupported |
-
-## Generating a live report
-
-Requires a running Sandbox Agent server. `--endpoint` defaults to `http://127.0.0.1:2468`.
-
-```bash
-sandbox-agent api agents report
-```
-
-<Note>
-    The live report reflects what the agent adapter returns for the current credentials. Some models may be gated by subscription (e.g. Claude's `opus` requires a paid plan) and will not appear in the report if the credentials don't have access.
-</Note>

docs/agent-sessions.mdx

@@ -21,10 +21,7 @@ const sdk = await SandboxAgent.connect({
 
 const session = await sdk.createSession({
   agent: "codex",
-  sessionInit: {
+  cwd: "/",
-    cwd: "/",
-    mcpServers: [],
-  },
 });
 
 console.log(session.id, session.agentSessionId);
@@ -54,6 +51,108 @@ await session.prompt([
 unsubscribe();
 ```
 
+### Event types
+
+Each event's `payload` contains a session update. The `sessionUpdate` field identifies the type.
+
+<AccordionGroup>
+<Accordion title="agent_message_chunk">
+Streamed text or content from the agent's response.
+
+```json
+{
+  "sessionUpdate": "agent_message_chunk",
+  "content": { "type": "text", "text": "Here's how the repository is structured..." }
+}
+```
+</Accordion>
+
+<Accordion title="agent_thought_chunk">
+Internal reasoning from the agent (chain-of-thought / extended thinking).
+
+```json
+{
+  "sessionUpdate": "agent_thought_chunk",
+  "content": { "type": "text", "text": "I should start by looking at the project structure..." }
+}
+```
+</Accordion>
+
+<Accordion title="user_message_chunk">
+Echo of the user's prompt being processed.
+
+```json
+{
+  "sessionUpdate": "user_message_chunk",
+  "content": { "type": "text", "text": "Summarize the repository structure." }
+}
+```
+</Accordion>
+
+<Accordion title="tool_call">
+The agent invoked a tool (file edit, terminal command, etc.).
+
+```json
+{
+  "sessionUpdate": "tool_call",
+  "toolCallId": "tc_abc123",
+  "title": "Read file",
+  "status": "in_progress",
+  "rawInput": { "path": "/src/index.ts" }
+}
+```
+</Accordion>
+
+<Accordion title="tool_call_update">
+Progress or result update for an in-progress tool call.
+
+```json
+{
+  "sessionUpdate": "tool_call_update",
+  "toolCallId": "tc_abc123",
+  "status": "completed",
+  "content": [{ "type": "text", "text": "import express from 'express';\n..." }]
+}
+```
+</Accordion>
+
+<Accordion title="plan">
+The agent's execution plan for the current task.
+
+```json
+{
+  "sessionUpdate": "plan",
+  "entries": [
+    { "content": "Read the project structure", "status": "completed" },
+    { "content": "Identify main entrypoints", "status": "in_progress" },
+    { "content": "Write summary", "status": "pending" }
+  ]
+}
+```
+</Accordion>
+
+<Accordion title="usage_update">
+Token usage metrics for the current turn.
+
+```json
+{
+  "sessionUpdate": "usage_update"
+}
+```
+</Accordion>
+
+<Accordion title="session_info_update">
+Session metadata changed (e.g. agent-generated title).
+
+```json
+{
+  "sessionUpdate": "session_info_update",
+  "title": "Repository structure analysis"
+}
+```
+</Accordion>
+</AccordionGroup>
+
 ## Fetch persisted event history
 
 ```ts

docs/agents/amp.mdx

@@ -0,0 +1,20 @@
+---
+title: "Amp"
+description: "Use Amp as a sandbox agent."
+---
+
+## Usage
+
+```typescript
+const session = await client.createSession({
+  agent: "amp",
+});
+```
+
+## Capabilities
+
+| Category | Values |
+|----------|--------|
+| **Models** | `amp-default` |
+| **Modes** | `default`, `bypass` |
+| **Thought levels** | Unsupported |

docs/agents/claude.mdx

@@ -0,0 +1,49 @@
+---
+title: "Claude"
+description: "Use Claude Code as a sandbox agent."
+---
+
+## Usage
+
+```typescript
+const session = await client.createSession({
+  agent: "claude",
+});
+```
+
+## Capabilities
+
+| Category | Values |
+|----------|--------|
+| **Models** | `default`, `sonnet`, `opus`, `haiku` |
+| **Modes** | `default`, `acceptEdits`, `plan`, `dontAsk`, `bypassPermissions` |
+| **Thought levels** | Unsupported |
+
+## Configuring effort level
+
+Claude does not support changing effort level after a session starts. Configure it in the filesystem before creating the session.
+
+```ts
+import { mkdir, writeFile } from "node:fs/promises";
+import path from "node:path";
+
+const cwd = "/path/to/workspace";
+await mkdir(path.join(cwd, ".claude"), { recursive: true });
+await writeFile(
+  path.join(cwd, ".claude", "settings.json"),
+  JSON.stringify({ effortLevel: "high" }, null, 2),
+);
+
+const session = await client.createSession({
+  agent: "claude",
+  cwd,
+});
+```
+
+<Accordion title="Supported settings file locations (highest precedence last)">
+
+1. `~/.claude/settings.json`
+2. `<session cwd>/.claude/settings.json`
+3. `<session cwd>/.claude/settings.local.json`
+
+</Accordion>

docs/agents/codex.mdx

@@ -0,0 +1,20 @@
+---
+title: "Codex"
+description: "Use OpenAI Codex as a sandbox agent."
+---
+
+## Usage
+
+```typescript
+const session = await client.createSession({
+  agent: "codex",
+});
+```
+
+## Capabilities
+
+| Category | Values |
+|----------|--------|
+| **Models** | `gpt-5.3-codex` (default), `gpt-5.3-codex-spark`, `gpt-5.2-codex`, `gpt-5.1-codex-max`, `gpt-5.2`, `gpt-5.1-codex-mini` |
+| **Modes** | `read-only` (default), `auto`, `full-access` |
+| **Thought levels** | `low`, `medium`, `high` (default), `xhigh` |

docs/agents/cursor.mdx

@@ -0,0 +1,34 @@
+---
+title: "Cursor"
+description: "Use Cursor as a sandbox agent."
+---
+
+## Usage
+
+```typescript
+const session = await client.createSession({
+  agent: "cursor",
+});
+```
+
+## Capabilities
+
+| Category | Values |
+|----------|--------|
+| **Models** | See below |
+| **Modes** | Unsupported |
+| **Thought levels** | Unsupported |
+
+<Accordion title="All models">
+
+| Group | Models |
+|-------|--------|
+| **Auto** | `auto` |
+| **Composer** | `composer-1.5`, `composer-1` |
+| **GPT-5.3 Codex** | `gpt-5.3-codex`, `gpt-5.3-codex-low`, `gpt-5.3-codex-high`, `gpt-5.3-codex-xhigh`, `gpt-5.3-codex-fast`, `gpt-5.3-codex-low-fast`, `gpt-5.3-codex-high-fast`, `gpt-5.3-codex-xhigh-fast` |
+| **GPT-5.2** | `gpt-5.2`, `gpt-5.2-high`, `gpt-5.2-codex`, `gpt-5.2-codex-low`, `gpt-5.2-codex-high`, `gpt-5.2-codex-xhigh`, `gpt-5.2-codex-fast`, `gpt-5.2-codex-low-fast`, `gpt-5.2-codex-high-fast`, `gpt-5.2-codex-xhigh-fast` |
+| **GPT-5.1** | `gpt-5.1-high`, `gpt-5.1-codex-max`, `gpt-5.1-codex-max-high` |
+| **Claude** | `opus-4.6-thinking` (default), `opus-4.6`, `opus-4.5`, `opus-4.5-thinking`, `sonnet-4.5`, `sonnet-4.5-thinking` |
+| **Other** | `gemini-3-pro`, `gemini-3-flash`, `grok` |
+
+</Accordion>

docs/agents/opencode.mdx

@@ -0,0 +1,31 @@
+---
+title: "OpenCode"
+description: "Use OpenCode as a sandbox agent."
+---
+
+## Usage
+
+```typescript
+const session = await client.createSession({
+  agent: "opencode",
+});
+```
+
+## Capabilities
+
+| Category | Values |
+|----------|--------|
+| **Models** | See below |
+| **Modes** | `build` (default), `plan` |
+| **Thought levels** | Unsupported |
+
+<Accordion title="All models">
+
+| Provider | Models |
+|----------|--------|
+| **Anthropic** | `anthropic/claude-3-5-haiku-20241022`, `anthropic/claude-3-5-haiku-latest`, `anthropic/claude-3-5-sonnet-20240620`, `anthropic/claude-3-5-sonnet-20241022`, `anthropic/claude-3-7-sonnet-20250219`, `anthropic/claude-3-7-sonnet-latest`, `anthropic/claude-3-haiku-20240307`, `anthropic/claude-3-opus-20240229`, `anthropic/claude-3-sonnet-20240229`, `anthropic/claude-haiku-4-5`, `anthropic/claude-haiku-4-5-20251001`, `anthropic/claude-opus-4-0`, `anthropic/claude-opus-4-1`, `anthropic/claude-opus-4-1-20250805`, `anthropic/claude-opus-4-20250514`, `anthropic/claude-opus-4-5`, `anthropic/claude-opus-4-5-20251101`, `anthropic/claude-opus-4-6`, `anthropic/claude-sonnet-4-0`, `anthropic/claude-sonnet-4-20250514`, `anthropic/claude-sonnet-4-5`, `anthropic/claude-sonnet-4-5-20250929` |
+| **OpenAI** | `openai/gpt-5.1-codex`, `openai/gpt-5.1-codex-max`, `openai/gpt-5.1-codex-mini`, `openai/gpt-5.2`, `openai/gpt-5.2-codex`, `openai/gpt-5.3-codex` |
+| **Cerebras** | `cerebras/gpt-oss-120b`, `cerebras/qwen-3-235b-a22b-instruct-2507`, `cerebras/zai-glm-4.7` |
+| **OpenCode Zen** | `opencode/big-pickle`, `opencode/claude-3-5-haiku`, `opencode/claude-haiku-4-5`, `opencode/claude-opus-4-1`, `opencode/claude-opus-4-5`, `opencode/claude-opus-4-6`, `opencode/claude-sonnet-4`, `opencode/claude-sonnet-4-5`, `opencode/gemini-3-flash`, `opencode/gemini-3-pro` (default), `opencode/glm-4.6`, `opencode/glm-4.7`, `opencode/gpt-5`, `opencode/gpt-5-codex`, `opencode/gpt-5-nano`, `opencode/gpt-5.1`, `opencode/gpt-5.1-codex`, `opencode/gpt-5.1-codex-max`, `opencode/gpt-5.1-codex-mini`, `opencode/gpt-5.2`, `opencode/gpt-5.2-codex`, `opencode/kimi-k2`, `opencode/kimi-k2-thinking`, `opencode/kimi-k2.5`, `opencode/kimi-k2.5-free`, `opencode/minimax-m2.1`, `opencode/minimax-m2.1-free`, `opencode/trinity-large-preview-free` |
+
+</Accordion>

docs/agents/pi.mdx

@@ -0,0 +1,20 @@
+---
+title: "Pi"
+description: "Use Pi as a sandbox agent."
+---
+
+## Usage
+
+```typescript
+const session = await client.createSession({
+  agent: "pi",
+});
+```
+
+## Capabilities
+
+| Category | Values |
+|----------|--------|
+| **Models** | `default` |
+| **Modes** | Unsupported |
+| **Thought levels** | Unsupported |

docs/architecture.mdx

@@ -1,64 +1,63 @@
 ---
 title: "Architecture"
-description: "How the client, sandbox, server, and agent fit together."
+description: "How the Sandbox Agent server, SDK, and agent processes fit together."
-icon: "microchip"
 ---
 
-Sandbox Agent runs as an HTTP server inside your sandbox. Your app talks to it remotely.
+Sandbox Agent is a lightweight HTTP server that runs **inside** a sandbox. It:
+
+- **Agent management**: Installs, spawns, and stops coding agent processes
+- **Sessions**: Routes prompts to agents and streams events back in real time
+- **Sandbox APIs**: Filesystem, process, and terminal access for the sandbox environment
 
 ## Components
 
-- `Your client`: your app code using the `sandbox-agent` SDK.
+```mermaid
-- `Sandbox`: isolated runtime (E2B, Daytona, Docker, etc.).
+flowchart LR
-- `Sandbox Agent server`: process inside the sandbox exposing HTTP transport.
+    CLIENT["Your App"]
-- `Agent`: Claude/Codex/OpenCode/Amp process managed by Sandbox Agent.
-
-```mermaid placement="top-right"
-  flowchart LR
-    CLIENT["Sandbox Agent SDK"]
-    SERVER["Sandbox Agent server"]
-    AGENT["Agent process"]
 
     subgraph SANDBOX["Sandbox"]
-      direction TB
+        direction TB
-      SERVER --> AGENT
+        SERVER["Sandbox Agent Server"]
+        AGENT["Agent Process<br/>(Claude, Codex, etc.)"]
+        SERVER --> AGENT
     end
 
-    CLIENT -->|HTTP| SERVER
+    CLIENT -->|"SDK (HTTP)"| SERVER
 ```
 
-## Suggested Topology
+- **Your app**: Uses the `sandbox-agent` TypeScript SDK to talk to the server over HTTP.
+- **Sandbox**: An isolated runtime (local process, Docker, E2B, Daytona, Vercel, Cloudflare).
+- **Sandbox Agent server**: A single binary inside the sandbox that manages agent lifecycles, routes prompts, streams events, and exposes filesystem/process/terminal APIs.
+- **Agent process**: A coding agent (Claude Code, Codex, etc.) spawned by the server. Each session maps to one agent process.
 
-Run the SDK on your backend, then call it from your frontend.
+## What `SandboxAgent.start()` does
 
-This extra hop is recommended because it keeps auth/token logic on the backend and makes persistence simpler.
+1. **Provision**: The provider creates a sandbox (starts a container, creates a VM, etc.)
+2. **Install**: The Sandbox Agent binary is installed inside the sandbox
+3. **Boot**: The server starts listening on an HTTP port
+4. **Health check**: The SDK waits for `/v1/health` to respond
+5. **Ready**: The SDK returns a connected client
 
-```mermaid placement="top-right"
+For the `local` provider, provisioning is a no-op and the server runs as a local subprocess.
-  flowchart LR
-    BROWSER["Browser"]
-    subgraph BACKEND["Your backend"]
-      direction TB
-      SDK["Sandbox Agent SDK"]
-    end
-    subgraph SANDBOX_SIMPLE["Sandbox"]
-      SERVER_SIMPLE["Sandbox Agent server"]
-    end
 
-    BROWSER --> BACKEND
+### Server recovery
-    BACKEND --> SDK --> SERVER_SIMPLE
+
+If the server process stops, the SDK automatically calls the provider's `ensureServer()` after 3 consecutive health-check failures. Most built-in providers implement this. Custom providers can add `ensureServer(sandboxId)` to their `SandboxProvider` object.
+
+## Server HTTP API
+
+See the [HTTP API reference](/api-reference) for the full list of server endpoints.
+
+## Agent installation
+
+Agents are installed lazily on first use. To avoid the cold-start delay, pre-install them:
+
+```bash
+sandbox-agent install-agent --all
 ```
 
-### Backend requirements
+The `rivetdev/sandbox-agent:0.4.2-full` Docker image ships with all agents pre-installed.
 
-Your backend layer needs to handle:
+## Production-ready agent orchestration
 
-- **Long-running connections**: prompts can take minutes.
+For production deployments, see [Orchestration Architecture](/orchestration-architecture) for recommended topology, backend requirements, and session persistence patterns.
-- **Session affinity**: follow-up messages must reach the same session.
-- **State between requests**: session metadata and event history must persist across requests.
-- **Graceful recovery**: sessions should resume after backend restarts.
-
-We recommend [Rivet](https://rivet.dev) over serverless because actors natively support the long-lived connections, session routing, and state persistence that agent workloads require.
-
-## Session persistence
-
-For storage driver options and replay behavior, see [Persisting Sessions](/session-persistence).

docs/cli.mdx

@@ -37,6 +37,36 @@ Notes:
 - Set `SANDBOX_AGENT_LOG_STDOUT=1` to force stdout/stderr logging.
 - Use `SANDBOX_AGENT_LOG_DIR` to override log directory.
 
+## install
+
+Install first-party runtime dependencies.
+
+### install desktop
+
+Install the Linux desktop runtime packages required by `/v1/desktop/*`.
+
+```bash
+sandbox-agent install desktop [OPTIONS]
+```
+
+| Option | Description |
+|--------|-------------|
+| `--yes` | Skip the confirmation prompt |
+| `--print-only` | Print the package-manager command without executing it |
+| `--package-manager <apt\|dnf\|apk>` | Override package-manager detection |
+| `--no-fonts` | Skip the default DejaVu font package |
+
+```bash
+sandbox-agent install desktop --yes
+sandbox-agent install desktop --print-only
+```
+
+Notes:
+
+- Supported on Linux only.
+- The command detects `apt`, `dnf`, or `apk`.
+- If the host is not already running as root, the command requires `sudo`.
+
 ## install-agent
 
 Install or reinstall a single agent, or every supported agent with `--all`.
@@ -59,6 +89,39 @@ sandbox-agent install-agent claude --reinstall
 sandbox-agent install-agent --all
 ```
 
+### Custom Pi implementation path
+
+If you use a forked/custom `pi` binary with `pi-acp`, you can override what executable gets launched.
+
+#### Option 1: explicit command override (recommended)
+
+Set `PI_ACP_PI_COMMAND` in the environment where `sandbox-agent` runs:
+
+```bash
+PI_ACP_PI_COMMAND=/absolute/path/to/your/pi-fork sandbox-agent server
+```
+
+This is forwarded to `pi-acp`, which uses it instead of looking up `pi` on `PATH`.
+
+#### Option 2: PATH override
+
+Put your custom `pi` first on `PATH` before starting `sandbox-agent`:
+
+```bash
+export PATH="/path/to/custom-pi-dir:$PATH"
+sandbox-agent server
+```
+
+#### Option 3: symlink override
+
+Point `pi` to your custom binary via symlink in a directory that is early on `PATH`:
+
+```bash
+ln -sf /absolute/path/to/your/pi-fork /usr/local/bin/pi
+```
+
+Then start `sandbox-agent` normally.
+
 ## opencode (experimental)
 
 Start/reuse daemon and run `opencode attach` against `/opencode`.
@@ -226,7 +289,7 @@ Example output:
 }
 ```
 
-See [Agent Capabilities](/agent-capabilities) for a full reference of supported models, modes, and thought levels per agent.
+See individual agent pages (e.g. [Claude](/agents/claude), [Codex](/agents/codex)) for supported models, modes, and thought levels.
 
 #### api agents install
 

docs/common-software.mdx

@@ -0,0 +1,560 @@
+---
+title: "Common Software"
+description: "Install browsers, languages, databases, and other tools inside the sandbox."
+sidebarTitle: "Common Software"
+icon: "box-open"
+---
+
+The sandbox runs a Debian/Ubuntu base image. You can install software with `apt-get` via the [Process API](/processes) or by customizing your Docker image. This page covers commonly needed packages and how to install them.
+
+## Browsers
+
+### Chromium
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "chromium", "chromium-sandbox"],
+});
+
+// Launch headless
+await sdk.runProcess({
+  command: "chromium",
+  args: ["--headless", "--no-sandbox", "--disable-gpu", "https://example.com"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","chromium","chromium-sandbox"]}'
+```
+</CodeGroup>
+
+<Note>
+Use `--no-sandbox` when running Chromium inside a container. The container itself provides isolation.
+</Note>
+
+### Firefox
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "firefox-esr"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","firefox-esr"]}'
+```
+</CodeGroup>
+
+### Playwright browsers
+
+Playwright bundles its own browser binaries. Install the Playwright CLI and let it download browsers for you.
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "npx",
+  args: ["playwright", "install", "--with-deps", "chromium"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"npx","args":["playwright","install","--with-deps","chromium"]}'
+```
+</CodeGroup>
+
+---
+
+## Languages and runtimes
+
+### Node.js
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "nodejs", "npm"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","nodejs","npm"]}'
+```
+</CodeGroup>
+
+For a specific version, use [nvm](https://github.com/nvm-sh/nvm):
+
+```ts TypeScript
+await sdk.runProcess({
+  command: "bash",
+  args: ["-c", "curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.1/install.sh | bash && . ~/.nvm/nvm.sh && nvm install 22"],
+});
+```
+
+### Python
+
+Python 3 is typically pre-installed. To add pip and common packages:
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "python3", "python3-pip", "python3-venv"],
+});
+
+await sdk.runProcess({
+  command: "pip3",
+  args: ["install", "numpy", "pandas", "matplotlib"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","python3","python3-pip","python3-venv"]}'
+
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"pip3","args":["install","numpy","pandas","matplotlib"]}'
+```
+</CodeGroup>
+
+### Go
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "bash",
+  args: ["-c", "curl -fsSL https://go.dev/dl/go1.23.6.linux-amd64.tar.gz | tar -C /usr/local -xz"],
+});
+
+// Add to PATH for subsequent commands
+await sdk.runProcess({
+  command: "bash",
+  args: ["-c", "export PATH=$PATH:/usr/local/go/bin && go version"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"bash","args":["-c","curl -fsSL https://go.dev/dl/go1.23.6.linux-amd64.tar.gz | tar -C /usr/local -xz"]}'
+```
+</CodeGroup>
+
+### Rust
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "bash",
+  args: ["-c", "curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"bash","args":["-c","curl --proto =https --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y"]}'
+```
+</CodeGroup>
+
+### Java (OpenJDK)
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "default-jdk"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","default-jdk"]}'
+```
+</CodeGroup>
+
+### Ruby
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "ruby-full"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","ruby-full"]}'
+```
+</CodeGroup>
+
+---
+
+## Databases
+
+### PostgreSQL
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "postgresql", "postgresql-client"],
+});
+
+// Start the service
+const proc = await sdk.createProcess({
+  command: "bash",
+  args: ["-c", "su - postgres -c 'pg_ctlcluster 15 main start'"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","postgresql","postgresql-client"]}'
+```
+</CodeGroup>
+
+### SQLite
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "sqlite3"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","sqlite3"]}'
+```
+</CodeGroup>
+
+### Redis
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "redis-server"],
+});
+
+const proc = await sdk.createProcess({
+  command: "redis-server",
+  args: ["--daemonize", "no"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","redis-server"]}'
+
+curl -X POST "http://127.0.0.1:2468/v1/processes" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"redis-server","args":["--daemonize","no"]}'
+```
+</CodeGroup>
+
+### MySQL / MariaDB
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "mariadb-server", "mariadb-client"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","mariadb-server","mariadb-client"]}'
+```
+</CodeGroup>
+
+---
+
+## Build tools
+
+### Essential build toolchain
+
+Most compiled software needs the standard build toolchain:
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "build-essential", "cmake", "pkg-config"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","build-essential","cmake","pkg-config"]}'
+```
+</CodeGroup>
+
+This installs `gcc`, `g++`, `make`, `cmake`, and related tools.
+
+---
+
+## Desktop applications
+
+These require the [Computer Use](/computer-use) desktop to be started first.
+
+### LibreOffice
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "libreoffice"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","libreoffice"]}'
+```
+</CodeGroup>
+
+### GIMP
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "gimp"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","gimp"]}'
+```
+</CodeGroup>
+
+### VLC
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "vlc"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","vlc"]}'
+```
+</CodeGroup>
+
+### VS Code (code-server)
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "bash",
+  args: ["-c", "curl -fsSL https://code-server.dev/install.sh | sh"],
+});
+
+const proc = await sdk.createProcess({
+  command: "code-server",
+  args: ["--bind-addr", "0.0.0.0:8080", "--auth", "none"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"bash","args":["-c","curl -fsSL https://code-server.dev/install.sh | sh"]}'
+
+curl -X POST "http://127.0.0.1:2468/v1/processes" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"code-server","args":["--bind-addr","0.0.0.0:8080","--auth","none"]}'
+```
+</CodeGroup>
+
+---
+
+## CLI tools
+
+### Git
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "git"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","git"]}'
+```
+</CodeGroup>
+
+### Docker
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "bash",
+  args: ["-c", "curl -fsSL https://get.docker.com | sh"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"bash","args":["-c","curl -fsSL https://get.docker.com | sh"]}'
+```
+</CodeGroup>
+
+### jq
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "jq"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","jq"]}'
+```
+</CodeGroup>
+
+### tmux
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "tmux"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","tmux"]}'
+```
+</CodeGroup>
+
+---
+
+## Media and graphics
+
+### FFmpeg
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "ffmpeg"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","ffmpeg"]}'
+```
+</CodeGroup>
+
+### ImageMagick
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "imagemagick"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","imagemagick"]}'
+```
+</CodeGroup>
+
+### Poppler (PDF utilities)
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "poppler-utils"],
+});
+
+// Convert PDF to images
+await sdk.runProcess({
+  command: "pdftoppm",
+  args: ["-png", "document.pdf", "output"],
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","poppler-utils"]}'
+```
+</CodeGroup>
+
+---
+
+## Pre-installing in a Docker image
+
+For production use, install software in your Dockerfile instead of at runtime. This avoids repeated downloads and makes startup faster.
+
+```dockerfile
+FROM ubuntu:22.04
+
+RUN apt-get update && apt-get install -y \
+    chromium \
+    firefox-esr \
+    nodejs npm \
+    python3 python3-pip \
+    git curl wget \
+    build-essential \
+    sqlite3 \
+    ffmpeg \
+    imagemagick \
+    jq \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN pip3 install numpy pandas matplotlib
+```
+
+See [Docker deployment](/deploy/docker) for how to use custom images with Sandbox Agent.

docs/computer-use.mdx

@@ -0,0 +1,859 @@
+---
+title: "Computer Use"
+description: "Control a virtual desktop inside the sandbox with mouse, keyboard, screenshots, recordings, and live streaming."
+sidebarTitle: "Computer Use"
+icon: "desktop"
+---
+
+Sandbox Agent provides a managed virtual desktop (Xvfb + openbox) that you can control programmatically. This is useful for browser automation, GUI testing, and AI computer-use workflows.
+
+## Start and stop
+
+<CodeGroup>
+```ts TypeScript
+import { SandboxAgent } from "sandbox-agent";
+
+const sdk = await SandboxAgent.connect({
+  baseUrl: "http://127.0.0.1:2468",
+});
+
+const status = await sdk.startDesktop({
+  width: 1920,
+  height: 1080,
+  dpi: 96,
+});
+
+console.log(status.state); // "active"
+console.log(status.display); // ":99"
+
+// When done
+await sdk.stopDesktop();
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/desktop/start" \
+  -H "Content-Type: application/json" \
+  -d '{"width":1920,"height":1080,"dpi":96}'
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/stop"
+```
+</CodeGroup>
+
+All fields in the start request are optional. Defaults are 1440x900 at 96 DPI.
+
+### Start request options
+
+| Field | Type | Default | Description |
+|-------|------|---------|-------------|
+| `width` | number | 1440 | Desktop width in pixels |
+| `height` | number | 900 | Desktop height in pixels |
+| `dpi` | number | 96 | Display DPI |
+| `displayNum` | number | 99 | Starting X display number. The runtime probes from this number upward to find an available display. |
+| `stateDir` | string | (auto) | Desktop state directory for home, logs, recordings |
+| `streamVideoCodec` | string | `"vp8"` | WebRTC video codec (`vp8`, `vp9`, `h264`) |
+| `streamAudioCodec` | string | `"opus"` | WebRTC audio codec (`opus`, `g722`) |
+| `streamFrameRate` | number | 30 | Streaming frame rate (1-60) |
+| `webrtcPortRange` | string | `"59050-59070"` | UDP port range for WebRTC media |
+| `recordingFps` | number | 30 | Default recording FPS when not specified in `startDesktopRecording` (1-60) |
+
+The streaming and recording options configure defaults for the desktop session. They take effect when streaming or recording is started later.
+
+<CodeGroup>
+```ts TypeScript
+const status = await sdk.startDesktop({
+  width: 1920,
+  height: 1080,
+  streamVideoCodec: "h264",
+  streamFrameRate: 60,
+  webrtcPortRange: "59100-59120",
+  recordingFps: 15,
+});
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/desktop/start" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "width": 1920,
+    "height": 1080,
+    "streamVideoCodec": "h264",
+    "streamFrameRate": 60,
+    "webrtcPortRange": "59100-59120",
+    "recordingFps": 15
+  }'
+```
+</CodeGroup>
+
+## Status
+
+<CodeGroup>
+```ts TypeScript
+const status = await sdk.getDesktopStatus();
+console.log(status.state); // "inactive" | "active" | "failed" | ...
+```
+
+```bash cURL
+curl "http://127.0.0.1:2468/v1/desktop/status"
+```
+</CodeGroup>
+
+## Screenshots
+
+Capture the full desktop or a specific region. Optionally include the cursor position.
+
+<CodeGroup>
+```ts TypeScript
+// Full screenshot (PNG by default)
+const png = await sdk.takeDesktopScreenshot();
+
+// JPEG at 70% quality, half scale
+const jpeg = await sdk.takeDesktopScreenshot({
+  format: "jpeg",
+  quality: 70,
+  scale: 0.5,
+});
+
+// Include cursor overlay
+const withCursor = await sdk.takeDesktopScreenshot({
+  showCursor: true,
+});
+
+// Region screenshot
+const region = await sdk.takeDesktopRegionScreenshot({
+  x: 100,
+  y: 100,
+  width: 400,
+  height: 300,
+});
+```
+
+```bash cURL
+curl "http://127.0.0.1:2468/v1/desktop/screenshot" --output screenshot.png
+
+curl "http://127.0.0.1:2468/v1/desktop/screenshot?format=jpeg&quality=70&scale=0.5" \
+  --output screenshot.jpg
+
+# Include cursor overlay
+curl "http://127.0.0.1:2468/v1/desktop/screenshot?show_cursor=true" \
+  --output with_cursor.png
+
+curl "http://127.0.0.1:2468/v1/desktop/screenshot/region?x=100&y=100&width=400&height=300" \
+  --output region.png
+```
+</CodeGroup>
+
+### Screenshot options
+
+| Param | Type | Default | Description |
+|-------|------|---------|-------------|
+| `format` | string | `"png"` | Output format: `png`, `jpeg`, or `webp` |
+| `quality` | number | 85 | Compression quality (1-100, JPEG/WebP only) |
+| `scale` | number | 1.0 | Scale factor (0.1-1.0) |
+| `showCursor` | boolean | `false` | Composite a crosshair at the cursor position |
+
+When `showCursor` is enabled, the cursor position is captured at the moment of the screenshot and a red crosshair is drawn at that location. This is useful for AI agents that need to see where the cursor is in the screenshot.
+
+## Mouse
+
+<CodeGroup>
+```ts TypeScript
+// Get current position
+const pos = await sdk.getDesktopMousePosition();
+console.log(pos.x, pos.y);
+
+// Move
+await sdk.moveDesktopMouse({ x: 500, y: 300 });
+
+// Click (left by default)
+await sdk.clickDesktop({ x: 500, y: 300 });
+
+// Right click
+await sdk.clickDesktop({ x: 500, y: 300, button: "right" });
+
+// Double click
+await sdk.clickDesktop({ x: 500, y: 300, clickCount: 2 });
+
+// Drag
+await sdk.dragDesktopMouse({
+  startX: 100, startY: 100,
+  endX: 400, endY: 400,
+});
+
+// Scroll
+await sdk.scrollDesktop({ x: 500, y: 300, deltaY: -3 });
+```
+
+```bash cURL
+curl "http://127.0.0.1:2468/v1/desktop/mouse/position"
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/mouse/click" \
+  -H "Content-Type: application/json" \
+  -d '{"x":500,"y":300}'
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/mouse/drag" \
+  -H "Content-Type: application/json" \
+  -d '{"startX":100,"startY":100,"endX":400,"endY":400}'
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/mouse/scroll" \
+  -H "Content-Type: application/json" \
+  -d '{"x":500,"y":300,"deltaY":-3}'
+```
+</CodeGroup>
+
+## Keyboard
+
+<CodeGroup>
+```ts TypeScript
+// Type text
+await sdk.typeDesktopText({ text: "Hello, world!" });
+
+// Press a key with modifiers
+await sdk.pressDesktopKey({
+  key: "c",
+  modifiers: { ctrl: true },
+});
+
+// Low-level key down/up
+await sdk.keyDownDesktop({ key: "Shift_L" });
+await sdk.keyUpDesktop({ key: "Shift_L" });
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/desktop/keyboard/type" \
+  -H "Content-Type: application/json" \
+  -d '{"text":"Hello, world!"}'
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/keyboard/press" \
+  -H "Content-Type: application/json" \
+  -d '{"key":"c","modifiers":{"ctrl":true}}'
+```
+</CodeGroup>
+
+## Clipboard
+
+Read and write the X11 clipboard programmatically.
+
+<CodeGroup>
+```ts TypeScript
+// Read clipboard
+const clipboard = await sdk.getDesktopClipboard();
+console.log(clipboard.text);
+
+// Read primary selection (mouse-selected text)
+const primary = await sdk.getDesktopClipboard({ selection: "primary" });
+
+// Write to clipboard
+await sdk.setDesktopClipboard({ text: "Pasted via API" });
+
+// Write to both clipboard and primary selection
+await sdk.setDesktopClipboard({
+  text: "Synced text",
+  selection: "both",
+});
+```
+
+```bash cURL
+curl "http://127.0.0.1:2468/v1/desktop/clipboard"
+
+curl "http://127.0.0.1:2468/v1/desktop/clipboard?selection=primary"
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/clipboard" \
+  -H "Content-Type: application/json" \
+  -d '{"text":"Pasted via API"}'
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/clipboard" \
+  -H "Content-Type: application/json" \
+  -d '{"text":"Synced text","selection":"both"}'
+```
+</CodeGroup>
+
+The `selection` parameter controls which X11 selection to read or write:
+
+| Value | Description |
+|-------|-------------|
+| `clipboard` (default) | The standard clipboard (Ctrl+C / Ctrl+V) |
+| `primary` | The primary selection (text selected with the mouse) |
+| `both` | Write to both clipboard and primary selection (write only) |
+
+## Display and windows
+
+<CodeGroup>
+```ts TypeScript
+const display = await sdk.getDesktopDisplayInfo();
+console.log(display.resolution); // { width: 1920, height: 1080, dpi: 96 }
+
+const { windows } = await sdk.listDesktopWindows();
+for (const win of windows) {
+  console.log(win.title, win.x, win.y, win.width, win.height);
+}
+```
+
+```bash cURL
+curl "http://127.0.0.1:2468/v1/desktop/display/info"
+
+curl "http://127.0.0.1:2468/v1/desktop/windows"
+```
+</CodeGroup>
+
+The windows endpoint filters out noise automatically: window manager internals (Openbox), windows with empty titles, and tiny helper windows (under 120x80) are excluded. The currently active/focused window is always included regardless of filters.
+
+### Focused window
+
+Get the currently focused window without listing all windows.
+
+<CodeGroup>
+```ts TypeScript
+const focused = await sdk.getDesktopFocusedWindow();
+console.log(focused.title, focused.id);
+```
+
+```bash cURL
+curl "http://127.0.0.1:2468/v1/desktop/windows/focused"
+```
+</CodeGroup>
+
+Returns 404 if no window currently has focus.
+
+### Window management
+
+Focus, move, and resize windows by their X11 window ID.
+
+<CodeGroup>
+```ts TypeScript
+const { windows } = await sdk.listDesktopWindows();
+const win = windows[0];
+
+// Bring window to foreground
+await sdk.focusDesktopWindow(win.id);
+
+// Move window
+await sdk.moveDesktopWindow(win.id, { x: 100, y: 50 });
+
+// Resize window
+await sdk.resizeDesktopWindow(win.id, { width: 1280, height: 720 });
+```
+
+```bash cURL
+# Focus a window
+curl -X POST "http://127.0.0.1:2468/v1/desktop/windows/12345/focus"
+
+# Move a window
+curl -X POST "http://127.0.0.1:2468/v1/desktop/windows/12345/move" \
+  -H "Content-Type: application/json" \
+  -d '{"x":100,"y":50}'
+
+# Resize a window
+curl -X POST "http://127.0.0.1:2468/v1/desktop/windows/12345/resize" \
+  -H "Content-Type: application/json" \
+  -d '{"width":1280,"height":720}'
+```
+</CodeGroup>
+
+All three endpoints return the updated window info so you can verify the operation took effect. The window manager may adjust the requested position or size.
+
+## App launching
+
+Launch applications or open files/URLs on the desktop without needing to shell out.
+
+<CodeGroup>
+```ts TypeScript
+// Launch an app by name
+const result = await sdk.launchDesktopApp({
+  app: "firefox",
+  args: ["--private"],
+});
+console.log(result.processId); // "proc_7"
+
+// Launch and wait for the window to appear
+const withWindow = await sdk.launchDesktopApp({
+  app: "xterm",
+  wait: true,
+});
+console.log(withWindow.windowId); // "12345" or null if timed out
+
+// Open a URL with the default handler
+const opened = await sdk.openDesktopTarget({
+  target: "https://example.com",
+});
+console.log(opened.processId);
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/desktop/launch" \
+  -H "Content-Type: application/json" \
+  -d '{"app":"firefox","args":["--private"]}'
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/launch" \
+  -H "Content-Type: application/json" \
+  -d '{"app":"xterm","wait":true}'
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/open" \
+  -H "Content-Type: application/json" \
+  -d '{"target":"https://example.com"}'
+```
+</CodeGroup>
+
+The returned `processId` can be used with the [Process API](/processes) to read logs (`GET /v1/processes/{id}/logs`) or stop the application (`POST /v1/processes/{id}/stop`).
+
+When `wait` is `true`, the API polls for up to 5 seconds for a window to appear. If the window appears, its ID is returned in `windowId`. If it times out, `windowId` is `null` but the process is still running.
+
+<Tip>
+**Launch/Open vs the Process API:** Both `launch` and `open` are convenience wrappers around the [Process API](/processes). They create managed processes (with `owner: "desktop"`) that you can inspect, log, and stop through the same Process endpoints. The difference is that `launch` validates the binary exists in PATH first and can optionally wait for a window to appear, while `open` delegates to the system default handler (`xdg-open`). Use the Process API directly when you need full control over command, environment, working directory, or restart policies.
+</Tip>
+
+## Recording
+
+Record the desktop to MP4.
+
+<CodeGroup>
+```ts TypeScript
+const recording = await sdk.startDesktopRecording({ fps: 30 });
+console.log(recording.id);
+
+// ... do things ...
+
+const stopped = await sdk.stopDesktopRecording();
+
+// List all recordings
+const { recordings } = await sdk.listDesktopRecordings();
+
+// Download
+const mp4 = await sdk.downloadDesktopRecording(recording.id);
+
+// Clean up
+await sdk.deleteDesktopRecording(recording.id);
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/desktop/recording/start" \
+  -H "Content-Type: application/json" \
+  -d '{"fps":30}'
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/recording/stop"
+
+curl "http://127.0.0.1:2468/v1/desktop/recordings"
+
+curl "http://127.0.0.1:2468/v1/desktop/recordings/rec_1/download" --output recording.mp4
+
+curl -X DELETE "http://127.0.0.1:2468/v1/desktop/recordings/rec_1"
+```
+</CodeGroup>
+
+## Desktop processes
+
+The desktop runtime manages several background processes (Xvfb, openbox, neko, ffmpeg). These are all registered with the general [Process API](/processes) under the `desktop` owner, so you can inspect logs, check status, and troubleshoot using the same tools you use for any other managed process.
+
+<CodeGroup>
+```ts TypeScript
+// List all processes, including desktop-owned ones
+const { processes } = await sdk.listProcesses();
+
+const desktopProcs = processes.filter((p) => p.owner === "desktop");
+for (const p of desktopProcs) {
+  console.log(p.id, p.command, p.status);
+}
+
+// Read logs from a specific desktop process
+const logs = await sdk.getProcessLogs(desktopProcs[0].id, { tail: 50 });
+for (const entry of logs.entries) {
+  console.log(entry.stream, atob(entry.data));
+}
+```
+
+```bash cURL
+# List all processes (desktop processes have owner: "desktop")
+curl "http://127.0.0.1:2468/v1/processes"
+
+# Get logs from a specific desktop process
+curl "http://127.0.0.1:2468/v1/processes/proc_1/logs?tail=50"
+```
+</CodeGroup>
+
+The desktop status endpoint also includes a summary of running processes:
+
+<CodeGroup>
+```ts TypeScript
+const status = await sdk.getDesktopStatus();
+for (const proc of status.processes) {
+  console.log(proc.name, proc.pid, proc.running);
+}
+```
+
+```bash cURL
+curl "http://127.0.0.1:2468/v1/desktop/status"
+# Response includes: processes: [{ name: "Xvfb", pid: 123, running: true }, ...]
+```
+</CodeGroup>
+
+| Process | Role | Restart policy |
+|---------|------|---------------|
+| Xvfb | Virtual X11 framebuffer | Auto-restart while desktop is active |
+| openbox | Window manager | Auto-restart while desktop is active |
+| neko | WebRTC streaming server (started by `startDesktopStream`) | No auto-restart |
+| ffmpeg | Screen recorder (started by `startDesktopRecording`) | No auto-restart |
+
+## Live streaming
+
+Start a WebRTC stream for real-time desktop viewing in a browser.
+
+<CodeGroup>
+```ts TypeScript
+await sdk.startDesktopStream();
+
+// Check stream status
+const status = await sdk.getDesktopStreamStatus();
+console.log(status.active); // true
+console.log(status.processId); // "proc_5"
+
+// Connect via the React DesktopViewer component or
+// use the WebSocket signaling endpoint directly
+// at ws://127.0.0.1:2468/v1/desktop/stream/signaling
+
+await sdk.stopDesktopStream();
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/desktop/stream/start"
+
+# Check stream status
+curl "http://127.0.0.1:2468/v1/desktop/stream/status"
+
+# Connect to ws://127.0.0.1:2468/v1/desktop/stream/signaling for WebRTC signaling
+
+curl -X POST "http://127.0.0.1:2468/v1/desktop/stream/stop"
+```
+</CodeGroup>
+
+For a drop-in React component, see [React Components](/react-components).
+
+## API reference
+
+### Endpoints
+
+| Method | Path | Description |
+|--------|------|-------------|
+| `POST` | `/v1/desktop/start` | Start the desktop runtime |
+| `POST` | `/v1/desktop/stop` | Stop the desktop runtime |
+| `GET` | `/v1/desktop/status` | Get desktop runtime status |
+| `GET` | `/v1/desktop/screenshot` | Capture full desktop screenshot |
+| `GET` | `/v1/desktop/screenshot/region` | Capture a region screenshot |
+| `GET` | `/v1/desktop/mouse/position` | Get current mouse position |
+| `POST` | `/v1/desktop/mouse/move` | Move the mouse |
+| `POST` | `/v1/desktop/mouse/click` | Click the mouse |
+| `POST` | `/v1/desktop/mouse/down` | Press mouse button down |
+| `POST` | `/v1/desktop/mouse/up` | Release mouse button |
+| `POST` | `/v1/desktop/mouse/drag` | Drag from one point to another |
+| `POST` | `/v1/desktop/mouse/scroll` | Scroll at a position |
+| `POST` | `/v1/desktop/keyboard/type` | Type text |
+| `POST` | `/v1/desktop/keyboard/press` | Press a key with optional modifiers |
+| `POST` | `/v1/desktop/keyboard/down` | Press a key down (hold) |
+| `POST` | `/v1/desktop/keyboard/up` | Release a key |
+| `GET` | `/v1/desktop/display/info` | Get display info |
+| `GET` | `/v1/desktop/windows` | List visible windows |
+| `GET` | `/v1/desktop/windows/focused` | Get focused window info |
+| `POST` | `/v1/desktop/windows/{id}/focus` | Focus a window |
+| `POST` | `/v1/desktop/windows/{id}/move` | Move a window |
+| `POST` | `/v1/desktop/windows/{id}/resize` | Resize a window |
+| `GET` | `/v1/desktop/clipboard` | Read clipboard contents |
+| `POST` | `/v1/desktop/clipboard` | Write to clipboard |
+| `POST` | `/v1/desktop/launch` | Launch an application |
+| `POST` | `/v1/desktop/open` | Open a file or URL |
+| `POST` | `/v1/desktop/recording/start` | Start recording |
+| `POST` | `/v1/desktop/recording/stop` | Stop recording |
+| `GET` | `/v1/desktop/recordings` | List recordings |
+| `GET` | `/v1/desktop/recordings/{id}` | Get recording metadata |
+| `GET` | `/v1/desktop/recordings/{id}/download` | Download recording |
+| `DELETE` | `/v1/desktop/recordings/{id}` | Delete recording |
+| `POST` | `/v1/desktop/stream/start` | Start WebRTC streaming |
+| `POST` | `/v1/desktop/stream/stop` | Stop WebRTC streaming |
+| `GET` | `/v1/desktop/stream/status` | Get stream status |
+| `GET` | `/v1/desktop/stream/signaling` | WebSocket for WebRTC signaling |
+
+### TypeScript SDK methods
+
+| Method | Returns | Description |
+|--------|---------|-------------|
+| `startDesktop(request?)` | `DesktopStatusResponse` | Start the desktop |
+| `stopDesktop()` | `DesktopStatusResponse` | Stop the desktop |
+| `getDesktopStatus()` | `DesktopStatusResponse` | Get desktop status |
+| `takeDesktopScreenshot(query?)` | `Uint8Array` | Capture screenshot |
+| `takeDesktopRegionScreenshot(query)` | `Uint8Array` | Capture region screenshot |
+| `getDesktopMousePosition()` | `DesktopMousePositionResponse` | Get mouse position |
+| `moveDesktopMouse(request)` | `DesktopMousePositionResponse` | Move mouse |
+| `clickDesktop(request)` | `DesktopMousePositionResponse` | Click mouse |
+| `mouseDownDesktop(request)` | `DesktopMousePositionResponse` | Mouse button down |
+| `mouseUpDesktop(request)` | `DesktopMousePositionResponse` | Mouse button up |
+| `dragDesktopMouse(request)` | `DesktopMousePositionResponse` | Drag mouse |
+| `scrollDesktop(request)` | `DesktopMousePositionResponse` | Scroll |
+| `typeDesktopText(request)` | `DesktopActionResponse` | Type text |
+| `pressDesktopKey(request)` | `DesktopActionResponse` | Press key |
+| `keyDownDesktop(request)` | `DesktopActionResponse` | Key down |
+| `keyUpDesktop(request)` | `DesktopActionResponse` | Key up |
+| `getDesktopDisplayInfo()` | `DesktopDisplayInfoResponse` | Get display info |
+| `listDesktopWindows()` | `DesktopWindowListResponse` | List windows |
+| `getDesktopFocusedWindow()` | `DesktopWindowInfo` | Get focused window |
+| `focusDesktopWindow(id)` | `DesktopWindowInfo` | Focus a window |
+| `moveDesktopWindow(id, request)` | `DesktopWindowInfo` | Move a window |
+| `resizeDesktopWindow(id, request)` | `DesktopWindowInfo` | Resize a window |
+| `getDesktopClipboard(query?)` | `DesktopClipboardResponse` | Read clipboard |
+| `setDesktopClipboard(request)` | `DesktopActionResponse` | Write clipboard |
+| `launchDesktopApp(request)` | `DesktopLaunchResponse` | Launch an app |
+| `openDesktopTarget(request)` | `DesktopOpenResponse` | Open file/URL |
+| `startDesktopRecording(request?)` | `DesktopRecordingInfo` | Start recording |
+| `stopDesktopRecording()` | `DesktopRecordingInfo` | Stop recording |
+| `listDesktopRecordings()` | `DesktopRecordingListResponse` | List recordings |
+| `getDesktopRecording(id)` | `DesktopRecordingInfo` | Get recording |
+| `downloadDesktopRecording(id)` | `Uint8Array` | Download recording |
+| `deleteDesktopRecording(id)` | `void` | Delete recording |
+| `startDesktopStream()` | `DesktopStreamStatusResponse` | Start streaming |
+| `stopDesktopStream()` | `DesktopStreamStatusResponse` | Stop streaming |
+| `getDesktopStreamStatus()` | `DesktopStreamStatusResponse` | Stream status |
+
+## Customizing the desktop environment
+
+The desktop runs inside the sandbox filesystem, so you can customize it using the [File System](/file-system) API before or after starting the desktop. The desktop HOME directory is located at `~/.local/state/sandbox-agent/desktop/home` (or `$XDG_STATE_HOME/sandbox-agent/desktop/home` if `XDG_STATE_HOME` is set).
+
+All configuration files below are written to paths relative to this HOME directory.
+
+### Window manager (openbox)
+
+The desktop uses [openbox](http://openbox.org/) as its window manager. You can customize its behavior, theme, and keyboard shortcuts by writing an `rc.xml` config file.
+
+<CodeGroup>
+```ts TypeScript
+const openboxConfig = `<?xml version="1.0" encoding="UTF-8"?>
+<openbox_config xmlns="http://openbox.org/3.4/rc">
+  <theme>
+    <name>Clearlooks</name>
+    <titleLayout>NLIMC</titleLayout>
+    <font place="ActiveWindow"><name>DejaVu Sans</name><size>10</size></font>
+  </theme>
+  <desktops><number>1</number></desktops>
+  <keyboard>
+    <keybind key="A-F4"><action name="Close"/></keybind>
+    <keybind key="A-Tab"><action name="NextWindow"/></keybind>
+  </keyboard>
+</openbox_config>`;
+
+await sdk.mkdirFs({ path: "~/.local/state/sandbox-agent/desktop/home/.config/openbox" });
+await sdk.writeFsFile(
+  { path: "~/.local/state/sandbox-agent/desktop/home/.config/openbox/rc.xml" },
+  openboxConfig,
+);
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/fs/mkdir?path=~/.local/state/sandbox-agent/desktop/home/.config/openbox"
+
+curl -X PUT "http://127.0.0.1:2468/v1/fs/file?path=~/.local/state/sandbox-agent/desktop/home/.config/openbox/rc.xml" \
+  -H "Content-Type: application/octet-stream" \
+  --data-binary @rc.xml
+```
+</CodeGroup>
+
+### Autostart programs
+
+Openbox runs scripts in `~/.config/openbox/autostart` on startup. Use this to launch applications, set the background, or configure the environment.
+
+<CodeGroup>
+```ts TypeScript
+const autostart = `#!/bin/sh
+# Set a solid background color
+xsetroot -solid "#1e1e2e" &
+
+# Launch a terminal
+xterm -geometry 120x40+50+50 &
+
+# Launch a browser
+firefox --no-remote &
+`;
+
+await sdk.mkdirFs({ path: "~/.local/state/sandbox-agent/desktop/home/.config/openbox" });
+await sdk.writeFsFile(
+  { path: "~/.local/state/sandbox-agent/desktop/home/.config/openbox/autostart" },
+  autostart,
+);
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/fs/mkdir?path=~/.local/state/sandbox-agent/desktop/home/.config/openbox"
+
+curl -X PUT "http://127.0.0.1:2468/v1/fs/file?path=~/.local/state/sandbox-agent/desktop/home/.config/openbox/autostart" \
+  -H "Content-Type: application/octet-stream" \
+  --data-binary @autostart.sh
+```
+</CodeGroup>
+
+<Note>
+The autostart script runs when openbox starts, which happens during `startDesktop()`. Write the autostart file before calling `startDesktop()` for it to take effect.
+</Note>
+
+### Background
+
+There is no wallpaper set by default (the background is the X root window default). You can set it using `xsetroot` in the autostart script (as shown above), or use `feh` if you need an image:
+
+<CodeGroup>
+```ts TypeScript
+// Upload a wallpaper image
+import fs from "node:fs";
+
+const wallpaper = await fs.promises.readFile("./wallpaper.png");
+await sdk.writeFsFile(
+  { path: "~/.local/state/sandbox-agent/desktop/home/wallpaper.png" },
+  wallpaper,
+);
+
+// Set the autostart to apply it
+const autostart = `#!/bin/sh
+feh --bg-fill ~/wallpaper.png &
+`;
+
+await sdk.mkdirFs({ path: "~/.local/state/sandbox-agent/desktop/home/.config/openbox" });
+await sdk.writeFsFile(
+  { path: "~/.local/state/sandbox-agent/desktop/home/.config/openbox/autostart" },
+  autostart,
+);
+```
+
+```bash cURL
+curl -X PUT "http://127.0.0.1:2468/v1/fs/file?path=~/.local/state/sandbox-agent/desktop/home/wallpaper.png" \
+  -H "Content-Type: application/octet-stream" \
+  --data-binary @wallpaper.png
+
+curl -X PUT "http://127.0.0.1:2468/v1/fs/file?path=~/.local/state/sandbox-agent/desktop/home/.config/openbox/autostart" \
+  -H "Content-Type: application/octet-stream" \
+  --data-binary @autostart.sh
+```
+</CodeGroup>
+
+<Note>
+`feh` is not installed by default. Install it via the [Process API](/processes) before starting the desktop: `await sdk.runProcess({ command: "apt-get", args: ["install", "-y", "feh"] })`.
+</Note>
+
+### Fonts
+
+Only `fonts-dejavu-core` is installed by default. To add more fonts, install them with your system package manager or copy font files into the sandbox:
+
+<CodeGroup>
+```ts TypeScript
+// Install a font package
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "fonts-noto", "fonts-liberation"],
+});
+
+// Or copy a custom font file
+import fs from "node:fs";
+
+const font = await fs.promises.readFile("./CustomFont.ttf");
+await sdk.mkdirFs({ path: "~/.local/state/sandbox-agent/desktop/home/.local/share/fonts" });
+await sdk.writeFsFile(
+  { path: "~/.local/state/sandbox-agent/desktop/home/.local/share/fonts/CustomFont.ttf" },
+  font,
+);
+
+// Rebuild the font cache
+await sdk.runProcess({ command: "fc-cache", args: ["-fv"] });
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","fonts-noto","fonts-liberation"]}'
+
+curl -X POST "http://127.0.0.1:2468/v1/fs/mkdir?path=~/.local/state/sandbox-agent/desktop/home/.local/share/fonts"
+
+curl -X PUT "http://127.0.0.1:2468/v1/fs/file?path=~/.local/state/sandbox-agent/desktop/home/.local/share/fonts/CustomFont.ttf" \
+  -H "Content-Type: application/octet-stream" \
+  --data-binary @CustomFont.ttf
+
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"fc-cache","args":["-fv"]}'
+```
+</CodeGroup>
+
+### Cursor theme
+
+<CodeGroup>
+```ts TypeScript
+await sdk.runProcess({
+  command: "apt-get",
+  args: ["install", "-y", "dmz-cursor-theme"],
+});
+
+const xresources = `Xcursor.theme: DMZ-White\nXcursor.size: 24\n`;
+await sdk.writeFsFile(
+  { path: "~/.local/state/sandbox-agent/desktop/home/.Xresources" },
+  xresources,
+);
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/processes/run" \
+  -H "Content-Type: application/json" \
+  -d '{"command":"apt-get","args":["install","-y","dmz-cursor-theme"]}'
+
+curl -X PUT "http://127.0.0.1:2468/v1/fs/file?path=~/.local/state/sandbox-agent/desktop/home/.Xresources" \
+  -H "Content-Type: application/octet-stream" \
+  --data-binary 'Xcursor.theme: DMZ-White\nXcursor.size: 24'
+```
+</CodeGroup>
+
+<Note>
+Run `xrdb -merge ~/.Xresources` (via the autostart or process API) after writing the file for changes to take effect.
+</Note>
+
+### Shell and terminal
+
+No terminal emulator or shell is launched by default. Add one to the openbox autostart:
+
+```sh
+# In ~/.config/openbox/autostart
+xterm -geometry 120x40+50+50 &
+```
+
+To use a different shell, set the `SHELL` environment variable in your Dockerfile or install your preferred shell and configure the terminal to use it.
+
+### GTK theme
+
+Applications using GTK will pick up settings from `~/.config/gtk-3.0/settings.ini`:
+
+<CodeGroup>
+```ts TypeScript
+const gtkSettings = `[Settings]
+gtk-theme-name=Adwaita
+gtk-icon-theme-name=Adwaita
+gtk-font-name=DejaVu Sans 10
+gtk-cursor-theme-name=DMZ-White
+gtk-cursor-theme-size=24
+`;
+
+await sdk.mkdirFs({ path: "~/.local/state/sandbox-agent/desktop/home/.config/gtk-3.0" });
+await sdk.writeFsFile(
+  { path: "~/.local/state/sandbox-agent/desktop/home/.config/gtk-3.0/settings.ini" },
+  gtkSettings,
+);
+```
+
+```bash cURL
+curl -X POST "http://127.0.0.1:2468/v1/fs/mkdir?path=~/.local/state/sandbox-agent/desktop/home/.config/gtk-3.0"
+
+curl -X PUT "http://127.0.0.1:2468/v1/fs/file?path=~/.local/state/sandbox-agent/desktop/home/.config/gtk-3.0/settings.ini" \
+  -H "Content-Type: application/octet-stream" \
+  --data-binary @settings.ini
+```
+</CodeGroup>
+
+### Summary of configuration paths
+
+All paths are relative to the desktop HOME directory (`~/.local/state/sandbox-agent/desktop/home`).
+
+| What | Path | Notes |
+|------|------|-------|
+| Openbox config | `.config/openbox/rc.xml` | Window manager theme, keybindings, behavior |
+| Autostart | `.config/openbox/autostart` | Shell script run on desktop start |
+| Custom fonts | `.local/share/fonts/` | TTF/OTF files, run `fc-cache -fv` after |
+| Cursor theme | `.Xresources` | Requires `xrdb -merge` to apply |
+| GTK 3 settings | `.config/gtk-3.0/settings.ini` | Theme, icons, fonts for GTK apps |
+| Wallpaper | Any path, referenced from autostart | Requires `feh` or similar tool |

docs/credentials.mdx

@@ -1,115 +0,0 @@
----
-title: "Credentials"
-description: "How Sandbox Agent discovers and uses provider credentials."
----
-
-Sandbox Agent discovers API credentials from environment variables and local agent config files.
-These credentials are passed through to underlying agent runtimes.
-
-## Credential sources
-
-Credentials are discovered in priority order.
-
-### Environment variables (highest priority)
-
-API keys first:
-
-| Variable | Provider |
-|----------|----------|
-| `ANTHROPIC_API_KEY` | Anthropic |
-| `CLAUDE_API_KEY` | Anthropic fallback |
-| `OPENAI_API_KEY` | OpenAI |
-| `CODEX_API_KEY` | OpenAI fallback |
-
-OAuth tokens (used when OAuth extraction is enabled):
-
-| Variable | Provider |
-|----------|----------|
-| `CLAUDE_CODE_OAUTH_TOKEN` | Anthropic |
-| `ANTHROPIC_AUTH_TOKEN` | Anthropic fallback |
-
-### Agent config files
-
-| Agent | Config path | Provider |
-|-------|-------------|----------|
-| Amp | `~/.amp/config.json` | Anthropic |
-| Claude Code | `~/.claude.json`, `~/.claude/.credentials.json` | Anthropic |
-| Codex | `~/.codex/auth.json` | OpenAI |
-| OpenCode | `~/.local/share/opencode/auth.json` | Anthropic/OpenAI |
-
-## Provider requirements by agent
-
-| Agent | Required provider |
-|-------|-------------------|
-| Claude Code | Anthropic |
-| Amp | Anthropic |
-| Codex | OpenAI |
-| OpenCode | Anthropic or OpenAI |
-| Mock | None |
-
-## Error handling behavior
-
-Credential extraction is best-effort:
-
-- Missing or malformed files are skipped.
-- Discovery continues to later sources.
-- Missing credentials mark providers unavailable instead of failing server startup.
-
-When prompting, Sandbox Agent does not pre-validate provider credentials. Agent-native authentication errors surface through session events/output.
-
-## Checking credential status
-
-### API
-
-`sdk.listAgents()` includes `credentialsAvailable` per agent.
-
-```json
-{
-  "agents": [
-    {
-      "id": "claude",
-      "installed": true,
-      "credentialsAvailable": true
-    },
-    {
-      "id": "codex",
-      "installed": true,
-      "credentialsAvailable": false
-    }
-  ]
-}
-```
-
-### TypeScript SDK
-
-```typescript
-const result = await sdk.listAgents();
-
-for (const agent of result.agents) {
-  console.log(`${agent.id}: ${agent.credentialsAvailable ? "authenticated" : "no credentials"}`);
-}
-```
-
-## Passing credentials explicitly
-
-Set environment variables before starting Sandbox Agent:
-
-```bash
-export ANTHROPIC_API_KEY=sk-ant-...
-export OPENAI_API_KEY=sk-...
-sandbox-agent daemon start
-```
-
-Or with SDK-managed local spawn:
-
-```typescript
-import { SandboxAgent } from "sandbox-agent";
-
-const sdk = await SandboxAgent.start({
-  spawn: {
-    env: {
-      ANTHROPIC_API_KEY: process.env.MY_ANTHROPIC_KEY,
-    },
-  },
-});
-```

docs/custom-tools.mdx

@@ -80,9 +80,7 @@ await sdk.setMcpConfig(
 
 const session = await sdk.createSession({
   agent: "claude",
-  sessionInit: {
+  cwd: "/workspace",
-    cwd: "/workspace",
-  },
 });
 
 await session.prompt([
@@ -145,9 +143,7 @@ await sdk.writeFsFile({ path: "/opt/skills/random-number/SKILL.md" }, skill);
 ```ts
 const session = await sdk.createSession({
   agent: "claude",
-  sessionInit: {
+  cwd: "/workspace",
-    cwd: "/workspace",
-  },
 });
 
 await session.prompt([

docs/deploy/boxlite.mdx

@@ -20,7 +20,7 @@ that BoxLite can load directly (BoxLite has its own image store separate from Do
 ```dockerfile
 FROM node:22-bookworm-slim
 RUN apt-get update && apt-get install -y curl ca-certificates && rm -rf /var/lib/apt/lists/*
-RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh
+RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh
 RUN sandbox-agent install-agent claude
 RUN sandbox-agent install-agent codex
 ```

docs/deploy/cloudflare.mdx

@@ -25,13 +25,44 @@ cd my-sandbox
 ```dockerfile
 FROM cloudflare/sandbox:0.7.0
 
-RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh
+RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh
 RUN sandbox-agent install-agent claude && sandbox-agent install-agent codex
 
 EXPOSE 8000
 ```
 
-## TypeScript example
+## TypeScript example (with provider)
+
+For standalone scripts, use the `cloudflare` provider:
+
+```bash
+npm install sandbox-agent@0.4.x @cloudflare/sandbox
+```
+
+```typescript
+import { SandboxAgent } from "sandbox-agent";
+import { cloudflare } from "sandbox-agent/cloudflare";
+
+const sdk = await SandboxAgent.start({
+  sandbox: cloudflare(),
+});
+
+try {
+  const session = await sdk.createSession({ agent: "codex" });
+  const response = await session.prompt([
+    { type: "text", text: "Summarize this repository" },
+  ]);
+  console.log(response.stopReason);
+} finally {
+  await sdk.destroySandbox();
+}
+```
+
+The `cloudflare` provider uses `containerFetch` under the hood, automatically stripping `AbortSignal` to avoid dropped streaming updates.
+
+## TypeScript example (Durable Objects)
+
+For Workers with Durable Objects, use `SandboxAgent.connect(...)` with a custom `fetch` backed by `sandbox.containerFetch(...)`:
 
 ```typescript
 import { getSandbox, type Sandbox } from "@cloudflare/sandbox";
@@ -109,7 +140,6 @@ app.all("*", (c) => c.env.ASSETS.fetch(c.req.raw));
 export default app;
 ```
 
-Create the SDK client inside the Worker using custom `fetch` backed by `sandbox.containerFetch(...)`.
 This keeps all Sandbox Agent calls inside the Cloudflare sandbox routing path and does not require a `baseUrl`.
 
 ## Troubleshooting streaming updates

docs/deploy/computesdk.mdx

@@ -1,160 +1,66 @@
 ---
 title: "ComputeSDK"
-description: "Deploy the daemon using ComputeSDK's provider-agnostic sandbox API."
+description: "Deploy Sandbox Agent using ComputeSDK's provider-agnostic sandbox API."
 ---
 
-[ComputeSDK](https://computesdk.com) provides a unified interface for managing sandboxes across multiple providers. Write once, deploy anywhere—switch providers by changing environment variables.
+[ComputeSDK](https://computesdk.com) provides a unified interface for managing sandboxes across multiple providers. Write once, deploy anywhere by changing environment variables.
 
 ## Prerequisites
 
 - `COMPUTESDK_API_KEY` from [console.computesdk.com](https://console.computesdk.com)
 - Provider API key (one of: `E2B_API_KEY`, `DAYTONA_API_KEY`, `VERCEL_TOKEN`, `MODAL_TOKEN_ID` + `MODAL_TOKEN_SECRET`, `BLAXEL_API_KEY`, `CSB_API_KEY`)
-- `ANTHROPIC_API_KEY` or `OPENAI_API_KEY` for the coding agents
+- `ANTHROPIC_API_KEY` or `OPENAI_API_KEY`
 
-## TypeScript Example
+## TypeScript example
+
+```bash
+npm install sandbox-agent@0.4.x computesdk
+```
 
 ```typescript
-import {
-  compute,
-  detectProvider,
-  getMissingEnvVars,
-  getProviderConfigFromEnv,
-  isProviderAuthComplete,
-  isValidProvider,
-  PROVIDER_NAMES,
-  type ExplicitComputeConfig,
-  type ProviderName,
-} from "computesdk";
 import { SandboxAgent } from "sandbox-agent";
+import { computesdk } from "sandbox-agent/computesdk";
 
-const PORT = 3000;
-const REQUEST_TIMEOUT_MS =
-  Number.parseInt(process.env.COMPUTESDK_TIMEOUT_MS || "", 10) || 120_000;
-
-/**
- * Detects and validates the provider to use.
- * Priority: COMPUTESDK_PROVIDER env var > auto-detection from API keys
- */
-function resolveProvider(): ProviderName {
-  const providerOverride = process.env.COMPUTESDK_PROVIDER;
-
-  if (providerOverride) {
-    if (!isValidProvider(providerOverride)) {
-      throw new Error(
-        `Unsupported provider "${providerOverride}". Supported: ${PROVIDER_NAMES.join(", ")}`
-      );
-    }
-    if (!isProviderAuthComplete(providerOverride)) {
-      const missing = getMissingEnvVars(providerOverride);
-      throw new Error(
-        `Missing credentials for "${providerOverride}". Set: ${missing.join(", ")}`
-      );
-    }
-    return providerOverride as ProviderName;
-  }
-
-  const detected = detectProvider();
-  if (!detected) {
-    throw new Error(
-      `No provider credentials found. Set one of: ${PROVIDER_NAMES.map((p) => getMissingEnvVars(p).join(", ")).join(" | ")}`
-    );
-  }
-  return detected as ProviderName;
-}
-
-function configureComputeSDK(): void {
-  const provider = resolveProvider();
-
-  const config: ExplicitComputeConfig = {
-    provider,
-    computesdkApiKey: process.env.COMPUTESDK_API_KEY,
-    requestTimeoutMs: REQUEST_TIMEOUT_MS,
-  };
-
-  // Add provider-specific config from environment
-  const providerConfig = getProviderConfigFromEnv(provider);
-  if (Object.keys(providerConfig).length > 0) {
-    (config as any)[provider] = providerConfig;
-  }
-
-  compute.setConfig(config);
-}
-
-configureComputeSDK();
-
-// Build environment variables to pass to sandbox
 const envs: Record<string, string> = {};
 if (process.env.ANTHROPIC_API_KEY) envs.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
 if (process.env.OPENAI_API_KEY) envs.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
 
-// Create sandbox
+const sdk = await SandboxAgent.start({
-const sandbox = await compute.sandbox.create({
+  sandbox: computesdk({
-  envs: Object.keys(envs).length > 0 ? envs : undefined,
+    create: {
+      envs,
+      image: process.env.COMPUTESDK_IMAGE,
+      templateId: process.env.COMPUTESDK_TEMPLATE_ID,
+    },
+  }),
 });
 
-// Helper to run commands with error handling
+try {
-const run = async (cmd: string, options?: { background?: boolean }) => {
+  const session = await sdk.createSession({ agent: "claude" });
-  const result = await sandbox.runCommand(cmd, options);
+  const response = await session.prompt([
-  if (typeof result?.exitCode === "number" && result.exitCode !== 0) {
+    { type: "text", text: "Summarize this repository" },
-    throw new Error(`Command failed: ${cmd} (exit ${result.exitCode})\n${result.stderr || ""}`);
+  ]);
-  }
+  console.log(response.stopReason);
-  return result;
+} finally {
-};
+  await sdk.destroySandbox();
-
-// Install sandbox-agent
-await run("curl -fsSL https://releases.rivet.dev/sandbox-agent/latest/install.sh | sh");
-
-// Install agents conditionally based on available API keys
-if (envs.ANTHROPIC_API_KEY) {
-  await run("sandbox-agent install-agent claude");
 }
-if (envs.OPENAI_API_KEY) {
-  await run("sandbox-agent install-agent codex");
-}
-
-// Start the server in the background
-await run(`sandbox-agent server --no-token --host 0.0.0.0 --port ${PORT}`, { background: true });
-
-// Get the public URL for the sandbox
-const baseUrl = await sandbox.getUrl({ port: PORT });
-
-// Wait for server to be ready
-const deadline = Date.now() + REQUEST_TIMEOUT_MS;
-while (Date.now() < deadline) {
-  try {
-    const response = await fetch(`${baseUrl}/v1/health`);
-    if (response.ok) {
-      const data = await response.json();
-      if (data?.status === "ok") break;
-    }
-  } catch {
-    // Server not ready yet
-  }
-  await new Promise((r) => setTimeout(r, 500));
-}
-
-// Connect to the server
-const client = await SandboxAgent.connect({ baseUrl });
-
-// Detect which agent to use based on available API keys
-const agent = envs.ANTHROPIC_API_KEY ? "claude" : "codex";
-
-// Create a session and start coding
-await client.createSession("my-session", { agent });
-
-await client.postMessage("my-session", {
-  message: "Summarize this repository",
-});
-
-for await (const event of client.streamEvents("my-session")) {
-  console.log(event.type, event.data);
-}
-
-// Cleanup
-await sandbox.destroy();
 ```
 
-## Supported Providers
+The `computesdk` provider handles sandbox creation, Sandbox Agent installation, agent setup, and server startup automatically. ComputeSDK routes to your configured provider behind the scenes.
+The `create` option now forwards the full ComputeSDK sandbox-create payload, including provider-specific fields such as `image` and `templateId` when the selected provider supports them.
+
+Before calling `SandboxAgent.start()`, configure ComputeSDK with your provider:
+
+```typescript
+import { compute } from "computesdk";
+
+compute.setConfig({
+  provider: "e2b", // or auto-detect via detectProvider()
+  computesdkApiKey: process.env.COMPUTESDK_API_KEY,
+});
+```
+
+## Supported providers
 
 ComputeSDK auto-detects your provider from environment variables:
 
@@ -169,46 +75,7 @@ ComputeSDK auto-detects your provider from environment variables:
 
 ## Notes
 
-- **Provider resolution order**: `COMPUTESDK_PROVIDER` env var takes priority, otherwise auto-detection from API keys.
+- **Provider resolution**: Set `COMPUTESDK_PROVIDER` to force a specific provider, or let ComputeSDK auto-detect from API keys.
-- **Conditional agent installation**: Only agents with available API keys are installed, reducing setup time.
-- **Command error handling**: The example validates exit codes and throws on failures for easier debugging.
 - `sandbox.runCommand(..., { background: true })` keeps the server running while your app continues.
 - `sandbox.getUrl({ port })` returns a public URL for the sandbox port.
-- Always destroy the sandbox when you are done to avoid leaking resources.
+- Always destroy the sandbox when done to avoid leaking resources.
-- If sandbox creation times out, set `COMPUTESDK_TIMEOUT_MS` to a higher value (default: 120000ms).
-
-## Explicit Provider Selection
-
-To force a specific provider instead of auto-detection, set the `COMPUTESDK_PROVIDER` environment variable:
-
-```bash
-export COMPUTESDK_PROVIDER=e2b
-```
-
-Or configure programmatically using `getProviderConfigFromEnv()`:
-
-```typescript
-import { compute, getProviderConfigFromEnv, type ExplicitComputeConfig } from "computesdk";
-
-const config: ExplicitComputeConfig = {
-  provider: "e2b",
-  computesdkApiKey: process.env.COMPUTESDK_API_KEY,
-  requestTimeoutMs: 120_000,
-};
-
-// Automatically populate provider-specific config from environment
-const providerConfig = getProviderConfigFromEnv("e2b");
-if (Object.keys(providerConfig).length > 0) {
-  (config as any).e2b = providerConfig;
-}
-
-compute.setConfig(config);
-```
-
-## Direct Mode (No ComputeSDK API Key)
-
-To bypass the ComputeSDK gateway and use provider SDKs directly, see the provider-specific examples:
-
-- [E2B](/deploy/e2b)
-- [Daytona](/deploy/daytona)
-- [Vercel](/deploy/vercel)

docs/deploy/daytona.mdx

@@ -15,40 +15,37 @@ See [Daytona network limits](https://www.daytona.io/docs/en/network-limits/).
 
 ## TypeScript example
 
-```typescript
+```bash
-import { Daytona } from "@daytonaio/sdk";
+npm install sandbox-agent@0.4.x @daytonaio/sdk
-import { SandboxAgent } from "sandbox-agent";
+```
 
-const daytona = new Daytona();
+```typescript
+import { SandboxAgent } from "sandbox-agent";
+import { daytona } from "sandbox-agent/daytona";
 
 const envVars: Record<string, string> = {};
 if (process.env.ANTHROPIC_API_KEY) envVars.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
 if (process.env.OPENAI_API_KEY) envVars.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
 
-const sandbox = await daytona.create({ envVars });
+const sdk = await SandboxAgent.start({
+  sandbox: daytona({
+    create: { envVars },
+  }),
+});
 
-await sandbox.process.executeCommand(
+try {
-  "curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh"
+  const session = await sdk.createSession({ agent: "claude" });
-);
+  const response = await session.prompt([
-
+    { type: "text", text: "Summarize this repository" },
-await sandbox.process.executeCommand("sandbox-agent install-agent claude");
+  ]);
-await sandbox.process.executeCommand("sandbox-agent install-agent codex");
+  console.log(response.stopReason);
-
+} finally {
-await sandbox.process.executeCommand(
+  await sdk.destroySandbox();
-  "nohup sandbox-agent server --no-token --host 0.0.0.0 --port 3000 >/tmp/sandbox-agent.log 2>&1 &"
+}
-);
-
-await new Promise((r) => setTimeout(r, 2000));
-
-const baseUrl = (await sandbox.getSignedPreviewUrl(3000, 4 * 60 * 60)).url;
-const sdk = await SandboxAgent.connect({ baseUrl });
-
-const session = await sdk.createSession({ agent: "claude" });
-await session.prompt([{ type: "text", text: "Summarize this repository" }]);
-
-await sandbox.delete();
 ```
 
+The `daytona` provider uses the `rivetdev/sandbox-agent:0.4.2-full` image by default and starts the server automatically.
+
 ## Using snapshots for faster startup
 
 ```typescript
@@ -64,7 +61,7 @@ if (!hasSnapshot) {
     name: SNAPSHOT,
     image: Image.base("ubuntu:22.04").runCommands(
       "apt-get update && apt-get install -y curl ca-certificates",
-      "curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh",
+      "curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh",
       "sandbox-agent install-agent claude",
       "sandbox-agent install-agent codex",
     ),

docs/deploy/docker.mdx

@@ -15,11 +15,32 @@ Run the published full image with all supported agents pre-installed:
 docker run --rm -p 3000:3000 \
   -e ANTHROPIC_API_KEY="$ANTHROPIC_API_KEY" \
   -e OPENAI_API_KEY="$OPENAI_API_KEY" \
-  rivetdev/sandbox-agent:0.3.1-full \
+  rivetdev/sandbox-agent:0.4.2-full \
   server --no-token --host 0.0.0.0 --port 3000
 ```
 
-The `0.3.1-full` tag pins the exact version. The moving `full` tag is also published for contributors who want the latest full image.
+The `0.4.2-full` tag pins the exact version. The moving `full` tag is also published for contributors who want the latest full image.
+
+If you also want the desktop API inside the container, install desktop dependencies before starting the server:
+
+```bash
+docker run --rm -p 3000:3000 \
+  -e ANTHROPIC_API_KEY="$ANTHROPIC_API_KEY" \
+  -e OPENAI_API_KEY="$OPENAI_API_KEY" \
+  node:22-bookworm-slim sh -c "\
+    apt-get update && \
+    DEBIAN_FRONTEND=noninteractive apt-get install -y curl ca-certificates bash libstdc++6 && \
+    rm -rf /var/lib/apt/lists/* && \
+    curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh && \
+    sandbox-agent install desktop --yes && \
+    sandbox-agent server --no-token --host 0.0.0.0 --port 3000"
+```
+
+In a Dockerfile:
+
+```dockerfile
+RUN sandbox-agent install desktop --yes
+```
 
 ## TypeScript with dockerode
 
@@ -31,7 +52,7 @@ const docker = new Docker();
 const PORT = 3000;
 
 const container = await docker.createContainer({
-  Image: "rivetdev/sandbox-agent:0.3.1-full",
+  Image: "rivetdev/sandbox-agent:0.4.2-full",
   Cmd: ["server", "--no-token", "--host", "0.0.0.0", "--port", `${PORT}`],
   Env: [
     `ANTHROPIC_API_KEY=${process.env.ANTHROPIC_API_KEY}`,
@@ -65,7 +86,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     bash ca-certificates curl git && \
     rm -rf /var/lib/apt/lists/*
 
-RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh && \
+RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh && \
     sandbox-agent install-agent --all
 
 RUN useradd -m -s /bin/bash sandbox

docs/deploy/e2b.mdx

@@ -10,43 +10,43 @@ description: "Deploy Sandbox Agent inside an E2B sandbox."
 
 ## TypeScript example
 
+```bash
+npm install sandbox-agent@0.4.x @e2b/code-interpreter
+```
+
 ```typescript
-import { Sandbox } from "@e2b/code-interpreter";
 import { SandboxAgent } from "sandbox-agent";
+import { e2b } from "sandbox-agent/e2b";
 
 const envs: Record<string, string> = {};
 if (process.env.ANTHROPIC_API_KEY) envs.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
 if (process.env.OPENAI_API_KEY) envs.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+const template = process.env.E2B_TEMPLATE;
 
-const sandbox = await Sandbox.create({ allowInternetAccess: true, envs });
+const sdk = await SandboxAgent.start({
-
+  sandbox: e2b({
-await sandbox.commands.run(
+    template,
-  "curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh"
+    create: { envs },
-);
+  }),
-
-await sandbox.commands.run("sandbox-agent install-agent claude");
-await sandbox.commands.run("sandbox-agent install-agent codex");
-
-await sandbox.commands.run(
-  "sandbox-agent server --no-token --host 0.0.0.0 --port 3000",
-  { background: true, timeoutMs: 0 }
-);
-
-const baseUrl = `https://${sandbox.getHost(3000)}`;
-const sdk = await SandboxAgent.connect({ baseUrl });
-
-const session = await sdk.createSession({ agent: "claude" });
-const off = session.onEvent((event) => {
-  console.log(event.sender, event.payload);
 });
 
-await session.prompt([{ type: "text", text: "Summarize this repository" }]);
+try {
-off();
+  const session = await sdk.createSession({ agent: "claude" });
-
+  const response = await session.prompt([
-await sandbox.kill();
+    { type: "text", text: "Summarize this repository" },
+  ]);
+  console.log(response.stopReason);
+} finally {
+  await sdk.destroySandbox();
+}
 ```
 
+The `e2b` provider handles sandbox creation, Sandbox Agent installation, agent setup, and server startup automatically. Sandboxes pause by default instead of being deleted, and reconnecting with the same `sandboxId` resumes them automatically.
+
+Pass `template` when you want to start from a custom E2B template alias or template ID. E2B base-image selection happens when you build the template, then `sandbox-agent/e2b` uses that template at sandbox creation time.
+
 ## Faster cold starts
 
 For faster startup, create a custom E2B template with Sandbox Agent and target agents pre-installed.
-See [E2B Custom Templates](https://e2b.dev/docs/sandbox-template).
+Build System 2.0 also lets you choose the template's base image in code.
+See [E2B Custom Templates](https://e2b.dev/docs/sandbox-template) and [E2B Base Images](https://e2b.dev/docs/template/base-image).

docs/deploy/foundry-self-hosting.mdx

@@ -1,153 +0,0 @@
----
-title: "Foundry Self-Hosting"
-description: "Environment, credentials, and deployment setup for Sandbox Agent Foundry auth, GitHub, and billing."
----
-
-This guide documents the deployment contract for the Foundry product surface: app auth, GitHub onboarding, repository import, and billing.
-
-It also covers the local-development bootstrap that uses `.env.development` only when `NODE_ENV=development`.
-
-## Local Development
-
-For backend local development, the Foundry backend now supports a development-only dotenv bootstrap:
-
-- It loads `.env.development.local` and `.env.development`
-- It does this **only** when `NODE_ENV=development`
-- It does **not** load dotenv files in production
-
-The example file lives at [`/.env.development.example`](https://github.com/rivet-dev/sandbox-agent/blob/main/.env.development.example).
-
-To use it locally:
-
-```bash
-cp .env.development.example .env.development
-```
-
-Run the backend with:
-
-```bash
-just foundry-backend-start
-```
-
-That recipe sets `NODE_ENV=development`, which enables the dotenv loader.
-
-### Local Defaults
-
-These values can be safely defaulted for local development:
-
-- `APP_URL=http://localhost:4173`
-- `BETTER_AUTH_URL=http://localhost:7741`
-- `BETTER_AUTH_SECRET=sandbox-agent-foundry-development-only-change-me`
-- `GITHUB_REDIRECT_URI=http://localhost:7741/v1/auth/callback/github`
-
-These should be treated as development-only values.
-
-## Production Environment
-
-For production or self-hosting, set these as real environment variables in your deployment platform. Do not rely on dotenv file loading.
-
-### App/Auth
-
-| Variable | Required | Notes |
-|---|---:|---|
-| `APP_URL` | Yes | Public frontend origin |
-| `BETTER_AUTH_URL` | Yes | Public auth base URL |
-| `BETTER_AUTH_SECRET` | Yes | Strong random secret for auth/session signing |
-
-### GitHub OAuth
-
-| Variable | Required | Notes |
-|---|---:|---|
-| `GITHUB_CLIENT_ID` | Yes | GitHub OAuth app client id |
-| `GITHUB_CLIENT_SECRET` | Yes | GitHub OAuth app client secret |
-| `GITHUB_REDIRECT_URI` | Yes | GitHub OAuth callback URL |
-
-Use GitHub OAuth for:
-
-- user sign-in
-- user identity
-- org selection
-- access to the signed-in user’s GitHub context
-
-## GitHub App
-
-If your Foundry deployment uses GitHub App-backed organization install and repo import, also configure:
-
-| Variable | Required | Notes |
-|---|---:|---|
-| `GITHUB_APP_ID` | Yes | GitHub App id |
-| `GITHUB_APP_CLIENT_ID` | Yes | GitHub App client id |
-| `GITHUB_APP_CLIENT_SECRET` | Yes | GitHub App client secret |
-| `GITHUB_APP_PRIVATE_KEY` | Yes | PEM private key for installation auth |
-
-For `.env.development` and `.env.development.local`, store `GITHUB_APP_PRIVATE_KEY` as a quoted single-line value with `\n` escapes instead of raw multi-line PEM text.
-
-Recommended GitHub App permissions:
-
-- Repository `Metadata: Read`
-- Repository `Contents: Read & Write`
-- Repository `Pull requests: Read & Write`
-- Repository `Checks: Read`
-- Repository `Commit statuses: Read`
-
-Set the webhook URL to `https://<your-backend-host>/v1/webhooks/github` and generate a webhook secret. Store the secret as `GITHUB_WEBHOOK_SECRET`.
-
-Recommended webhook subscriptions:
-
-- `installation`
-- `installation_repositories`
-- `pull_request`
-- `pull_request_review`
-- `pull_request_review_comment`
-- `push`
-- `create`
-- `delete`
-- `check_suite`
-- `check_run`
-- `status`
-
-Use the GitHub App for:
-
-- installation/reconnect state
-- org repo import
-- repository sync
-- PR creation and updates
-
-Use GitHub OAuth for:
-
-- who the user is
-- which orgs they can choose
-
-## Stripe
-
-For live billing, configure:
-
-| Variable | Required | Notes |
-|---|---:|---|
-| `STRIPE_SECRET_KEY` | Yes | Server-side Stripe secret key |
-| `STRIPE_PUBLISHABLE_KEY` | Yes | Client-side Stripe publishable key |
-| `STRIPE_WEBHOOK_SECRET` | Yes | Signing secret for billing webhooks |
-| `STRIPE_PRICE_TEAM` | Yes | Stripe price id for the Team plan checkout session |
-
-Stripe should own:
-
-- hosted checkout
-- billing portal
-- subscription status
-- invoice history
-- webhook-driven state sync
-
-## Mock Invariant
-
-Foundry’s mock client path should continue to work end to end even when the real auth/GitHub/Stripe path exists.
-
-That includes:
-
-- sign-in
-- org selection/import
-- settings
-- billing UI
-- workspace/task/session flow
-- seat accrual
-
-Use mock mode for deterministic UI review and local product development. Use the real env-backed path for integration and self-hosting.

docs/deploy/local.mdx

@@ -9,7 +9,7 @@ For local development, run Sandbox Agent directly on your machine.
 
 ```bash
 # Install
-curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh
+curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh
 
 # Run
 sandbox-agent server --no-token --host 127.0.0.1 --port 2468
@@ -20,24 +20,27 @@ Or with npm/Bun:
 <Tabs>
   <Tab title="npx">
     ```bash
-    npx @sandbox-agent/cli@0.3.x server --no-token --host 127.0.0.1 --port 2468
+    npx @sandbox-agent/cli@0.4.x server --no-token --host 127.0.0.1 --port 2468
     ```
   </Tab>
   <Tab title="bunx">
     ```bash
-    bunx @sandbox-agent/cli@0.3.x server --no-token --host 127.0.0.1 --port 2468
+    bunx @sandbox-agent/cli@0.4.x server --no-token --host 127.0.0.1 --port 2468
     ```
   </Tab>
 </Tabs>
 
 ## With the TypeScript SDK
 
-The SDK can spawn and manage the server as a subprocess:
+The SDK can spawn and manage the server as a subprocess using the `local` provider:
 
 ```typescript
 import { SandboxAgent } from "sandbox-agent";
+import { local } from "sandbox-agent/local";
 
-const sdk = await SandboxAgent.start();
+const sdk = await SandboxAgent.start({
+  sandbox: local(),
+});
 
 const session = await sdk.createSession({
   agent: "claude",
@@ -47,7 +50,21 @@ await session.prompt([
   { type: "text", text: "Summarize this repository." },
 ]);
 
-await sdk.dispose();
+await sdk.destroySandbox();
 ```
 
 This starts the server on an available local port and connects automatically.
+
+Pass options to customize the local provider:
+
+```typescript
+const sdk = await SandboxAgent.start({
+  sandbox: local({
+    port: 3000,
+    log: "inherit",
+    env: {
+      ANTHROPIC_API_KEY: process.env.MY_ANTHROPIC_KEY,
+    },
+  }),
+});
+```

docs/deploy/modal.mdx

@@ -0,0 +1,55 @@
+---
+title: "Modal"
+description: "Deploy Sandbox Agent inside a Modal sandbox."
+---
+
+## Prerequisites
+
+- `MODAL_TOKEN_ID` and `MODAL_TOKEN_SECRET` from [modal.com/settings](https://modal.com/settings)
+- `ANTHROPIC_API_KEY` or `OPENAI_API_KEY`
+
+## TypeScript example
+
+```bash
+npm install sandbox-agent@0.4.x modal
+```
+
+```typescript
+import { SandboxAgent } from "sandbox-agent";
+import { modal } from "sandbox-agent/modal";
+
+const secrets: Record<string, string> = {};
+if (process.env.ANTHROPIC_API_KEY) secrets.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
+if (process.env.OPENAI_API_KEY) secrets.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+const baseImage = process.env.MODAL_BASE_IMAGE ?? "node:22-slim";
+
+const sdk = await SandboxAgent.start({
+  sandbox: modal({
+    image: baseImage,
+    create: { secrets },
+  }),
+});
+
+try {
+  const session = await sdk.createSession({ agent: "claude" });
+  const response = await session.prompt([
+    { type: "text", text: "Summarize this repository" },
+  ]);
+  console.log(response.stopReason);
+} finally {
+  await sdk.destroySandbox();
+}
+```
+
+The `modal` provider handles app creation, image building, sandbox provisioning, agent installation, server startup, and tunnel networking automatically.
+Set `image` to change the base Docker image before Sandbox Agent and its agent binaries are layered on top. You can also pass a prebuilt Modal `Image` object.
+
+## Faster cold starts
+
+Modal caches image layers, so the Dockerfile commands that install `curl` and `sandbox-agent` only run on the first build. Subsequent sandbox creates reuse the cached image.
+
+## Notes
+
+- Modal sandboxes use [gVisor](https://gvisor.dev/) for strong isolation.
+- Ports are exposed via encrypted tunnels (`encryptedPorts`). The provider uses `sb.tunnels()` to get the public HTTPS URL.
+- Environment variables (API keys) are passed as Modal [Secrets](https://modal.com/docs/guide/secrets) for security.

docs/deploy/vercel.mdx

@@ -10,52 +10,40 @@ description: "Deploy Sandbox Agent inside a Vercel Sandbox."
 
 ## TypeScript example
 
-```typescript
+```bash
-import { Sandbox } from "@vercel/sandbox";
+npm install sandbox-agent@0.4.x @vercel/sandbox
-import { SandboxAgent } from "sandbox-agent";
-
-const envs: Record<string, string> = {};
-if (process.env.ANTHROPIC_API_KEY) envs.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
-if (process.env.OPENAI_API_KEY) envs.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
-
-const sandbox = await Sandbox.create({
-  runtime: "node24",
-  ports: [3000],
-});
-
-const run = async (cmd: string, args: string[] = []) => {
-  const result = await sandbox.runCommand({ cmd, args, env: envs });
-  if (result.exitCode !== 0) {
-    throw new Error(`Command failed: ${cmd} ${args.join(" ")}`);
-  }
-};
-
-await run("sh", ["-c", "curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh"]);
-await run("sandbox-agent", ["install-agent", "claude"]);
-await run("sandbox-agent", ["install-agent", "codex"]);
-
-await sandbox.runCommand({
-  cmd: "sandbox-agent",
-  args: ["server", "--no-token", "--host", "0.0.0.0", "--port", "3000"],
-  env: envs,
-  detached: true,
-});
-
-const baseUrl = sandbox.domain(3000);
-const sdk = await SandboxAgent.connect({ baseUrl });
-
-const session = await sdk.createSession({ agent: "claude" });
-
-const off = session.onEvent((event) => {
-  console.log(event.sender, event.payload);
-});
-
-await session.prompt([{ type: "text", text: "Summarize this repository" }]);
-off();
-
-await sandbox.stop();
 ```
 
+```typescript
+import { SandboxAgent } from "sandbox-agent";
+import { vercel } from "sandbox-agent/vercel";
+
+const env: Record<string, string> = {};
+if (process.env.ANTHROPIC_API_KEY) env.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
+if (process.env.OPENAI_API_KEY) env.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+
+const sdk = await SandboxAgent.start({
+  sandbox: vercel({
+    create: {
+      runtime: "node24",
+      env,
+    },
+  }),
+});
+
+try {
+  const session = await sdk.createSession({ agent: "claude" });
+  const response = await session.prompt([
+    { type: "text", text: "Summarize this repository" },
+  ]);
+  console.log(response.stopReason);
+} finally {
+  await sdk.destroySandbox();
+}
+```
+
+The `vercel` provider handles sandbox creation, Sandbox Agent installation, agent setup, and server startup automatically.
+
 ## Authentication
 
 Vercel Sandboxes support OIDC token auth (recommended) and access-token auth.

docs/docs.json

@@ -1,6 +1,6 @@
 {
   "$schema": "https://mintlify.com/docs.json",
-  "theme": "willow",
+  "theme": "mint",
   "name": "Sandbox Agent SDK",
   "appearance": {
     "default": "dark",
@@ -8,8 +8,8 @@
   },
   "colors": {
     "primary": "#ff4f00",
-    "light": "#ff4f00",
+    "light": "#ff6a2a",
-    "dark": "#ff4f00"
+    "dark": "#cc3f00"
   },
   "favicon": "/favicon.svg",
   "logo": {
@@ -25,17 +25,13 @@
   },
   "navbar": {
     "links": [
-      {
-        "label": "Gigacode",
-        "icon": "terminal",
-        "href": "https://github.com/rivet-dev/sandbox-agent/tree/main/gigacode"
-      },
       {
         "label": "Discord",
         "icon": "discord",
         "href": "https://discord.gg/auCecybynK"
       },
       {
+        "label": "GitHub",
         "type": "github",
         "href": "https://github.com/rivet-dev/sandbox-agent"
       }
@@ -51,46 +47,55 @@
             "pages": [
               "quickstart",
               "sdk-overview",
+              "llm-credentials",
               "react-components",
               {
                 "group": "Deploy",
                 "icon": "server",
                 "pages": [
                   "deploy/local",
-                  "deploy/computesdk",
                   "deploy/e2b",
                   "deploy/daytona",
                   "deploy/vercel",
                   "deploy/cloudflare",
                   "deploy/docker",
-                  "deploy/boxlite"
+                  "deploy/modal",
+                  "deploy/boxlite",
+                  "deploy/computesdk"
                 ]
               }
             ]
           },
           {
             "group": "Agent",
-            "pages": ["agent-sessions", "attachments", "skills-config", "mcp-config", "custom-tools"]
+            "pages": [
+              "agent-sessions",
+              {
+                "group": "Agents",
+                "icon": "robot",
+                "pages": ["agents/claude", "agents/codex", "agents/opencode", "agents/cursor", "agents/amp", "agents/pi"]
+              },
+              "attachments",
+              "skills-config",
+              "mcp-config",
+              "custom-tools"
+            ]
           },
           {
             "group": "System",
-            "pages": ["file-system", "processes"]
+            "pages": ["file-system", "processes", "computer-use", "common-software"]
-          },
-          {
-            "group": "Orchestration",
-            "pages": ["architecture", "session-persistence", "observability", "multiplayer", "security"]
           },
           {
             "group": "Reference",
             "pages": [
-              "agent-capabilities",
+              "troubleshooting",
+              "architecture",
               "cli",
               "inspector",
               "opencode-compatibility",
               {
                 "group": "More",
                 "pages": [
-                  "credentials",
                   "daemon",
                   "cors",
                   "session-restoration",
@@ -115,5 +120,11 @@
         ]
       }
     ]
-  }
+  },
+  "__removed": [
+	  {
+		"group": "Orchestration",
+		"pages": ["orchestration-architecture", "session-persistence", "observability", "multiplayer", "security"]
+	  }
+  ]
 }

docs/gigacode.mdx

@@ -1,6 +0,0 @@
----
-title: Gigacode
-url: "https://github.com/rivet-dev/sandbox-agent/tree/main/gigacode"
----
-
-

docs/inspector.mdx

@@ -35,6 +35,7 @@ console.log(url);
 - Prompt testing
 - Request/response debugging
 - Interactive permission prompts (approve, always-allow, or reject tool-use requests)
+- Desktop panel for status, remediation, start/stop, and screenshot refresh
 - Process management (create, stop, kill, delete, view logs)
 - Interactive PTY terminal for tty processes
 - One-shot command execution
@@ -50,3 +51,16 @@ console.log(url);
 The Inspector includes an embedded Ghostty-based terminal for interactive tty
 processes. The UI uses the SDK's high-level `connectProcessTerminal(...)`
 wrapper via the shared `@sandbox-agent/react` `ProcessTerminal` component.
+
+## Desktop panel
+
+The `Desktop` panel shows the current desktop runtime state, missing dependencies,
+the suggested install command, last error details, process/log paths, and the
+latest captured screenshot.
+
+Use it to:
+
+- Check whether desktop dependencies are installed
+- Start or stop the managed desktop runtime
+- Refresh desktop status
+- Capture a fresh screenshot on demand

docs/llm-credentials.mdx

@@ -0,0 +1,250 @@
+---
+title: "LLM Credentials"
+description: "Strategies for providing LLM provider credentials to agents."
+icon: "key"
+---
+
+Sandbox Agent needs LLM provider credentials (Anthropic, OpenAI, etc.) to run agent sessions.
+
+## Configuration
+
+Pass credentials via `spawn.env` when starting a sandbox. Each call to `SandboxAgent.start()` can use different credentials:
+
+```typescript
+import { SandboxAgent } from "sandbox-agent";
+
+const sdk = await SandboxAgent.start({
+  spawn: {
+    env: {
+      ANTHROPIC_API_KEY: "sk-ant-...",
+      OPENAI_API_KEY: "sk-...",
+    },
+  },
+});
+```
+
+Each agent requires credentials from a specific provider. Sandbox Agent checks environment variables (including those passed via `spawn.env`) and host config files:
+
+| Agent | Provider | Environment variables | Config files |
+|-------|----------|----------------------|--------------|
+| Claude Code | Anthropic | `ANTHROPIC_API_KEY`, `CLAUDE_API_KEY` | `~/.claude.json`, `~/.claude/.credentials.json` |
+| Amp | Anthropic | `ANTHROPIC_API_KEY`, `CLAUDE_API_KEY` | `~/.amp/config.json` |
+| Codex | OpenAI | `OPENAI_API_KEY`, `CODEX_API_KEY` | `~/.codex/auth.json` |
+| OpenCode | Anthropic or OpenAI | `ANTHROPIC_API_KEY`, `OPENAI_API_KEY` | `~/.local/share/opencode/auth.json` |
+| Mock | None | - | - |
+
+## Credential strategies
+
+LLM credentials are passed into the sandbox as environment variables. The agent and everything inside the sandbox has access to the token, so it's important to choose the right strategy for how you provision and scope these credentials.
+
+| Strategy | Who pays | Cost attribution | Best for |
+|----------|----------|-----------------|----------|
+| **Per-tenant gateway** (recommended) | Your organization, billed back per tenant | Per-tenant keys with budgets | Multi-tenant SaaS, usage-based billing |
+| **Bring your own key** | Each user (usage-based) | Per-user by default | Dev environments, internal tools |
+| **Shared API key** | Your organization | None (single bill) | Single-tenant apps, internal platforms |
+| **Personal subscription** | Each user (existing subscription) | Per-user by default | Local dev, internal tools where users have Claude or Codex subscriptions |
+
+### Per-tenant gateway (recommended)
+
+Route LLM traffic through a gateway that mints per-tenant API keys, each with its own spend tracking and budget limits.
+
+```mermaid
+graph LR
+    B[Your Backend] -->|tenant key| S[Sandbox]
+    S -->|LLM requests| G[Gateway]
+    G -->|scoped key| P[LLM Provider]
+```
+
+Your backend issues a scoped key per tenant, then passes it to the sandbox. This is the typical pattern when using sandbox providers (E2B, Daytona, Docker).
+
+```typescript expandable
+import { SandboxAgent } from "sandbox-agent";
+
+async function createTenantSandbox(tenantId: string) {
+  // Issue a scoped key for this tenant via OpenRouter
+  const res = await fetch("https://openrouter.ai/api/v1/keys", {
+    method: "POST",
+    headers: {
+      Authorization: `Bearer ${process.env.OPENROUTER_PROVISIONING_KEY}`,
+      "Content-Type": "application/json",
+    },
+    body: JSON.stringify({
+      name: `tenant-${tenantId}`,
+      limit: 50,
+      limitResetType: "monthly",
+    }),
+  });
+  const { key } = await res.json();
+
+  // Start a sandbox with the tenant's scoped key
+  const sdk = await SandboxAgent.start({
+    spawn: {
+      env: {
+        OPENAI_API_KEY: key, // OpenRouter uses OpenAI-compatible endpoints
+      },
+    },
+  });
+
+  const session = await sdk.createSession({
+    agent: "claude",
+    sessionInit: { cwd: "/workspace" },
+  });
+
+  return { sdk, session };
+}
+```
+
+#### Security
+
+Recommended for multi-tenant applications. Each tenant gets a scoped key with its own budget, so exfiltration only exposes that tenant's allowance.
+
+#### Use cases
+
+- **Multi-tenant SaaS**: per-tenant spend tracking and budget limits
+- **Production apps**: exposed to end users who need isolated credentials
+- **Usage-based billing**: each tenant pays for their own consumption
+
+#### Choosing a gateway
+
+<AccordionGroup>
+
+<Accordion title="OpenRouter provisioned keys" icon="cloud">
+
+Managed service, zero infrastructure. [OpenRouter](https://openrouter.ai/docs/features/provisioning-api-keys) provides per-tenant API keys with spend tracking and budget limits via their Provisioning API. Pass the tenant key to Sandbox Agent as `OPENAI_API_KEY` (OpenRouter uses OpenAI-compatible endpoints).
+
+```bash
+# Create a key for a tenant with a $50/month budget
+curl https://openrouter.ai/api/v1/keys \
+  -H "Authorization: Bearer $PROVISIONING_KEY" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "name": "tenant-acme",
+    "limit": 50,
+    "limitResetType": "monthly"
+  }'
+```
+
+Easiest to set up but not open-source. See [OpenRouter pricing](https://openrouter.ai/docs/framework/pricing) for details.
+
+</Accordion>
+
+<Accordion title="LiteLLM proxy" icon="server">
+
+Self-hosted, open-source (MIT). [LiteLLM](https://github.com/BerriAI/litellm) is an OpenAI-compatible proxy with hierarchical budgets (org, team, user, key), virtual keys, and spend tracking. Requires Python + PostgreSQL.
+
+```bash
+# Create a team (tenant) with a $500 budget
+curl http://litellm:4000/team/new \
+  -H "Authorization: Bearer $LITELLM_MASTER_KEY" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "team_alias": "tenant-acme",
+    "max_budget": 500
+  }'
+
+# Generate a key for that team
+curl http://litellm:4000/key/generate \
+  -H "Authorization: Bearer $LITELLM_MASTER_KEY" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "team_id": "team-abc123",
+    "max_budget": 100
+  }'
+```
+
+Full control with no vendor lock-in. Organization-level features require an enterprise license.
+
+</Accordion>
+
+<Accordion title="Portkey gateway" icon="code-branch">
+
+Self-hosted, open-source (Apache 2.0). [Portkey](https://github.com/Portkey-AI/gateway) is a lightweight OpenAI-compatible gateway supporting 200+ providers. Single binary, no database required. Create virtual keys with per-tenant budget limits and pass them to Sandbox Agent.
+
+Lightest operational footprint of the self-hosted options. Observability and analytics require the managed platform or your own tooling.
+
+</Accordion>
+
+</AccordionGroup>
+
+To bill tenants for LLM usage, use [Stripe token billing](https://docs.stripe.com/billing/token-billing) (integrates natively with OpenRouter) or query your gateway's spend API and feed usage into your billing system.
+
+### Bring your own key
+
+Each user provides their own API key. Users are billed directly by the LLM provider with no additional infrastructure needed.
+
+Pass the user's key via `spawn.env`:
+
+```typescript
+const sdk = await SandboxAgent.start({
+  spawn: {
+    env: {
+      ANTHROPIC_API_KEY: userProvidedKey,
+    },
+  },
+});
+```
+
+#### Security
+
+API keys are typically long-lived. The key is visible to the agent and anything running inside the sandbox, so exfiltration is possible. This is usually acceptable for developer-facing tools where the user owns the key.
+
+#### Use cases
+
+- **Developer tools**: each user manages their own API key
+- **Internal platforms**: users already have LLM provider accounts
+- **Per-user billing**: no extra infrastructure needed
+
+### Shared credentials
+
+A single organization-wide API key is used for all sessions. All token usage appears on one bill with no per-user or per-tenant cost attribution.
+
+```typescript
+const sdk = await SandboxAgent.start({
+  spawn: {
+    env: {
+      ANTHROPIC_API_KEY: process.env.ORG_ANTHROPIC_KEY!,
+      OPENAI_API_KEY: process.env.ORG_OPENAI_KEY!,
+    },
+  },
+});
+```
+
+If you need to track or limit spend per tenant, use a per-tenant gateway instead.
+
+#### Security
+
+Not recommended for anything other than internal tooling. A single exfiltrated key exposes your organization's entire LLM budget. If you need org-paid credentials for external users, use a per-tenant gateway with scoped keys instead.
+
+#### Use cases
+
+- **Single-tenant apps**: small number of users, one bill
+- **Prototyping**: cost attribution not needed yet
+- **Simplicity over security**: acceptable when exfiltration risk is low
+
+### Personal subscription
+
+If the user is signed into Claude Code or Codex on the host machine, Sandbox Agent automatically picks up their OAuth tokens. No configuration is needed.
+
+#### Remote sandboxes
+
+Extract credentials locally and pass them to a remote sandbox via `spawn.env`:
+
+```bash
+$ sandbox-agent credentials extract-env
+ANTHROPIC_API_KEY=sk-ant-...
+CLAUDE_API_KEY=sk-ant-...
+OPENAI_API_KEY=sk-...
+CODEX_API_KEY=sk-...
+```
+
+Use `-e` to prefix with `export` for shell sourcing.
+
+#### Security
+
+Personal subscriptions use OAuth tokens with a limited lifespan. These are the same credentials used when running an agent normally on the host. If a token is exfiltrated from the sandbox, the exposure window is short.
+
+#### Use cases
+
+- **Local development**: users are already signed into Claude Code or Codex
+- **Internal tools**: every user has their own subscription
+- **Prototyping**: no key management needed

docs/mcp-config.mdx

@@ -27,9 +27,7 @@ await sdk.setMcpConfig(
 // Create a session using the configured MCP servers
 const session = await sdk.createSession({
   agent: "claude",
-  sessionInit: {
+  cwd: "/workspace",
-    cwd: "/workspace",
-  },
 });
 
 await session.prompt([

docs/multiplayer.mdx

@@ -20,8 +20,40 @@ Use [actor keys](https://rivet.dev/docs/actors/keys) to map each workspace to on
 
 ```ts Actor (server)
 import { actor, setup } from "rivetkit";
-import { SandboxAgent } from "sandbox-agent";
+import { SandboxAgent, type SessionPersistDriver, type SessionRecord, type SessionEvent, type ListPageRequest, type ListPage, type ListEventsRequest } from "sandbox-agent";
-import { RivetSessionPersistDriver, type RivetPersistState } from "@sandbox-agent/persist-rivet";
+
+interface RivetPersistData { sessions: Record<string, SessionRecord>; events: Record<string, SessionEvent[]>; }
+type RivetPersistState = { _sandboxAgentPersist: RivetPersistData };
+
+class RivetSessionPersistDriver implements SessionPersistDriver {
+  private readonly stateKey: string;
+  private readonly ctx: { state: Record<string, unknown> };
+  constructor(ctx: { state: Record<string, unknown> }, options: { stateKey?: string } = {}) {
+    this.ctx = ctx;
+    this.stateKey = options.stateKey ?? "_sandboxAgentPersist";
+    if (!this.ctx.state[this.stateKey]) {
+      this.ctx.state[this.stateKey] = { sessions: {}, events: {} };
+    }
+  }
+  private get data(): RivetPersistData { return this.ctx.state[this.stateKey] as RivetPersistData; }
+  async getSession(id: string) { const s = this.data.sessions[id]; return s ? { ...s } : undefined; }
+  async listSessions(request: ListPageRequest = {}): Promise<ListPage<SessionRecord>> {
+    const sorted = Object.values(this.data.sessions).sort((a, b) => a.createdAt - b.createdAt || a.id.localeCompare(b.id));
+    const offset = Number(request.cursor ?? 0);
+    const limit = request.limit ?? 100;
+    const slice = sorted.slice(offset, offset + limit);
+    return { items: slice, nextCursor: offset + slice.length < sorted.length ? String(offset + slice.length) : undefined };
+  }
+  async updateSession(session: SessionRecord) { this.data.sessions[session.id] = { ...session }; if (!this.data.events[session.id]) this.data.events[session.id] = []; }
+  async listEvents(request: ListEventsRequest): Promise<ListPage<SessionEvent>> {
+    const all = [...(this.data.events[request.sessionId] ?? [])].sort((a, b) => a.eventIndex - b.eventIndex || a.id.localeCompare(b.id));
+    const offset = Number(request.cursor ?? 0);
+    const limit = request.limit ?? 100;
+    const slice = all.slice(offset, offset + limit);
+    return { items: slice, nextCursor: offset + slice.length < all.length ? String(offset + slice.length) : undefined };
+  }
+  async insertEvent(sessionId: string, event: SessionEvent) { const events = this.data.events[sessionId] ?? []; events.push({ ...event, payload: JSON.parse(JSON.stringify(event.payload)) }); this.data.events[sessionId] = events; }
+}
 
 type WorkspaceState = RivetPersistState & {
   sandboxId: string;
@@ -111,5 +143,5 @@ await conn.prompt({
 ## Notes
 
 - Keep sandbox calls actor-only. Browser clients should not call Sandbox Agent directly.
-- Use `@sandbox-agent/persist-rivet` so session history persists in actor state.
+- Copy the Rivet persist driver from the example above into your project so session history persists in actor state.
 - For client connection patterns, see [Rivet JavaScript client](https://rivet.dev/docs/clients/javascript).

docs/orchestration-architecture.mdx

@@ -0,0 +1,43 @@
+---
+title: "Orchestration Architecture"
+description: "Production topology, backend requirements, and session persistence."
+icon: "sitemap"
+---
+
+This page covers production topology and backend requirements. Read [Architecture](/architecture) first for an overview of how the server, SDK, and agent processes fit together.
+
+## Suggested Topology
+
+Run the SDK on your backend, then call it from your frontend.
+
+This extra hop is recommended because it keeps auth/token logic on the backend and makes persistence simpler.
+
+```mermaid placement="top-right"
+  flowchart LR
+    BROWSER["Browser"]
+    subgraph BACKEND["Your backend"]
+      direction TB
+      SDK["Sandbox Agent SDK"]
+    end
+    subgraph SANDBOX_SIMPLE["Sandbox"]
+      SERVER_SIMPLE["Sandbox Agent server"]
+    end
+
+    BROWSER --> BACKEND
+    BACKEND --> SDK --> SERVER_SIMPLE
+```
+
+### Backend requirements
+
+Your backend layer needs to handle:
+
+- **Long-running connections**: prompts can take minutes.
+- **Session affinity**: follow-up messages must reach the same session.
+- **State between requests**: session metadata and event history must persist across requests.
+- **Graceful recovery**: sessions should resume after backend restarts.
+
+We recommend [Rivet](https://rivet.dev) over serverless because actors natively support the long-lived connections, session routing, and state persistence that agent workloads require.
+
+## Session persistence
+
+For storage driver options and replay behavior, see [Persisting Sessions](/session-persistence).

docs/pi-support-plan.md

@@ -1,210 +0,0 @@
-# Pi Agent Support Plan (pi-mono)
-
-## Implementation Status Update
-
-- Runtime selection now supports two internal modes:
-  - `PerSession` (default for unknown/non-allowlisted Pi capabilities)
-  - `Shared` (allowlist-only compatibility path)
-- Pi sessions now use per-session process isolation by default, enabling true concurrent Pi sessions in Inspector and API clients.
-- Shared Pi server code remains available and is used only when capability checks allow multiplexing.
-- Session termination for per-session Pi mode hard-kills the underlying Pi process and clears queued prompts/pending waiters.
-- In-session concurrent sends are serialized with an unbounded daemon-side FIFO queue per session.
-
-## Investigation Summary
-
-### Pi CLI modes and RPC protocol
-- Pi supports multiple modes including interactive, print/JSON output, RPC, and SDK usage. JSON mode outputs a stream of JSON events suitable for parsing, and RPC mode is intended for programmatic control over stdin/stdout.
-- RPC mode is started with `pi --mode rpc` and supports options like `--provider`, `--model`, `--no-session`, and `--session-dir`.
-- The RPC protocol is newline-delimited JSON over stdin/stdout:
-  - Commands are JSON objects written to stdin.
-  - Responses are JSON objects with `type: "response"` and optional `id`.
-  - Events are JSON objects without `id`.
-- `prompt` can include images using `ImageContent` (base64 or URL) alongside text.
-- JSON/print mode (`pi -p` or `pi --print --mode json`) produces JSONL for non-interactive parsing and can resume sessions with a token.
-
-### RPC commands
-RPC commands listed in `rpc.md` include:
-- `new_session`, `get_state`, `list_sessions`, `delete_session`, `rename_session`, `clear_session`
-- `prompt`, `queue_message`, `abort`, `get_queued_messages`
-
-### RPC event types
-RPC events listed in `rpc.md` include:
-- `agent_start`, `agent_end`
-- `turn_start`, `turn_end`
-- `message_start`, `message_update`, `message_end`
-- `tool_execution_start`, `tool_execution_update`, `tool_execution_end`
-- `auto_compaction`, `auto_retry`, `hook_error`
-
-`message_update` uses `assistantMessageEvent` deltas such as:
-- `start`, `text_start`, `text_delta`, `text_end`
-- `thinking_start`, `thinking_delta`, `thinking_end`
-- `toolcall_start`, `toolcall_delta`, `toolcall_end`
-- `toolcall_args_start`, `toolcall_args_delta`, `toolcall_args_end`
-- `done`, `error`
-
-`tool_execution_update` includes `partialResult`, which is described as accumulated output so far.
-
-### Schema source locations (pi-mono)
-RPC types are documented as living in:
-- `packages/ai/src/types.ts` (Model types)
-- `packages/agent/src/types.ts` (AgentResponse types)
-- `packages/coding-agent/src/core/messages.ts` (message types)
-- `packages/coding-agent/src/modes/rpc/rpc-types.ts` (RPC protocol types)
-
-### Distribution assets
-Pi releases provide platform-specific binaries such as:
-- `pi-darwin-arm64`, `pi-darwin-x64`
-- `pi-linux-arm64`, `pi-linux-x64`
-- `pi-win-x64.zip`
-
-## Integration Decisions
-- Follow the OpenCode pattern: a shared long-running process (stdio RPC) with session multiplexing.
-- Primary integration path is RPC streaming (`pi --mode rpc`).
-- JSON/print mode is a fallback only (diagnostics or non-interactive runs).
-- Create sessions via `new_session`; store the returned `sessionId` as `native_session_id`.
-- Use `get_state` as a re-sync path after server restarts.
-- Use `prompt` for send-message, with optional image content.
-- Convert Pi events into universal events; emit daemon synthetic `session.started` on session creation and `session.ended` only on errors/termination.
-
-## Implementation Plan
-
-### 1) Agent Identity + Capabilities
-Files:
-- `server/packages/agent-management/src/agents.rs`
-- `server/packages/sandbox-agent/src/router.rs`
-- `docs/cli.mdx`, `docs/conversion.mdx`, `docs/session-transcript-schema.mdx`
-- `README.md`, `frontend/packages/website/src/components/FAQ.tsx`
-
-Tasks:
-- Add `AgentId::Pi` with string/binary name `"pi"` and parsing rules.
-- Add Pi to `all_agents()` and agent lists.
-- Define `AgentCapabilities` for Pi:
-  - `tool_calls=true`, `tool_results=true`
-  - `text_messages=true`, `streaming_deltas=true`, `item_started=true`
-  - `reasoning=true` (from `thinking_*` deltas)
-  - `images=true` (ImageContent in `prompt`)
-  - `permissions=false`, `questions=false`, `mcp_tools=false`
-  - `shared_process=true`, `session_lifecycle=false` (no native session events)
-  - `error_events=true` (hook_error)
-  - `command_execution=false`, `file_changes=false`, `file_attachments=false`
-
-### 2) Installer and Binary Resolution
-Files:
-- `server/packages/agent-management/src/agents.rs`
-
-Tasks:
-- Add `install_pi()` that:
-  - Downloads the correct release asset per platform (`pi-<platform>`).
-  - Handles `.zip` on Windows and raw binaries elsewhere.
-  - Marks binary executable.
-- Add Pi to `AgentManager::install`, `is_installed`, `version`.
-- Version detection: try `--version`, `version`, `-V`.
-
-### 3) Schema Extraction for Pi
-Files:
-- `resources/agent-schemas/src/pi.ts` (new)
-- `resources/agent-schemas/src/index.ts`
-- `resources/agent-schemas/artifacts/json-schema/pi.json`
-- `server/packages/extracted-agent-schemas/build.rs`
-- `server/packages/extracted-agent-schemas/src/lib.rs`
-
-Tasks:
-- Implement `extractPiSchema()`:
-  - Download pi-mono sources (zip/tarball) into a temp dir.
-  - Use `ts-json-schema-generator` against `packages/coding-agent/src/modes/rpc/rpc-types.ts`.
-  - Include dependent files per `rpc.md` (ai/types, agent/types, core/messages).
-  - Extract `RpcEvent`, `RpcResponse`, `RpcCommand` unions (exact type names from source).
-- Add fallback schema if remote fetch fails (minimal union with event/response fields).
-- Wire pi into extractor index and artifact generation.
-
-### 4) Universal Schema Conversion (Pi -> Universal)
-Files:
-- `server/packages/universal-agent-schema/src/agents/pi.rs` (new)
-- `server/packages/universal-agent-schema/src/agents/mod.rs`
-- `server/packages/universal-agent-schema/src/lib.rs`
-- `server/packages/sandbox-agent/src/router.rs`
-
-Mapping rules:
-- `message_start` -> `item.started` (kind=message, role=assistant, native_item_id=messageId)
-- `message_update`:
-  - `text_*` -> `item.delta` (assistant text delta)
-  - `thinking_*` -> `item.delta` with `ContentPart::Reasoning` (visibility=Private)
-  - `toolcall_*` and `toolcall_args_*` -> ignore for now (tool_execution_* is authoritative)
-  - `error` -> `item.completed` with `ItemStatus::Failed` (if no later message_end)
-- `message_end` -> `item.completed` (finalize assistant message)
-- `tool_execution_start` -> `item.started` (kind=tool_call, ContentPart::ToolCall)
-- `tool_execution_update` -> `item.delta` for a synthetic tool_result item:
-  - Maintain a per-toolCallId buffer to compute delta from accumulated `partialResult`.
-- `tool_execution_end` -> `item.completed` (kind=tool_result, output from `result.content`)
-  - If `isError=true`, set item status to failed.
-- `agent_start`, `turn_start`, `turn_end`, `agent_end`, `auto_compaction`, `auto_retry`, `hook_error`:
-  - Map to `ItemKind::Status` with a label like `pi.agent_start`, `pi.auto_retry`, etc.
-  - Do not emit `session.ended` for these events.
-- If event parsing fails, emit `agent.unparsed` (source=daemon, synthetic=true) and fail tests.
-
-### 5) Shared RPC Server Integration
-Files:
-- `server/packages/sandbox-agent/src/router.rs`
-
-Tasks:
-- Add a new managed stdio server type for Pi, similar to Codex:
-  - Create `PiServer` struct with:
-    - stdin sender
-    - pending request map keyed by request id
-    - per-session native session id mapping
-  - Extend `ManagedServerKind` to include Pi.
-  - Add `ensure_pi_server()` and `spawn_pi_server()` using `pi --mode rpc`.
-  - Add a `handle_pi_server_output()` loop to parse stdout lines into events/responses.
-- Session creation:
-  - On `create_session`, ensure Pi server is running, send `new_session`, store sessionId.
-  - Register session with `server_manager.register_session` for native mapping.
-- Sending messages:
-  - Use `prompt` command; include sessionId and optional images.
-  - Emit synthetic `item.started` only if Pi does not emit `message_start`.
-
-### 6) Router + Streaming Path Changes
-Files:
-- `server/packages/sandbox-agent/src/router.rs`
-
-Tasks:
-- Add Pi handling to:
-  - `create_session` (new_session)
-  - `send_message` (prompt)
-  - `parse_agent_line` (Pi event conversion)
-  - `agent_modes` (default to `default` unless Pi exposes a mode list)
-  - `agent_supports_resume` (true if Pi supports session resume)
-
-### 7) Tests
-Files:
-- `server/packages/sandbox-agent/tests/...`
-- `server/packages/universal-agent-schema/tests/...` (if present)
-
-Tasks:
-- Unit tests for conversion:
-  - `message_start/update/end` -> item.started/delta/completed
-  - `tool_execution_*` -> tool call/result mapping with partialResult delta
-  - failure -> agent.unparsed
-- Integration tests:
-  - Start Pi RPC server, create session, send prompt, stream events.
-  - Validate `native_session_id` mapping and event ordering.
-- Update HTTP/SSE test coverage to include Pi agent if relevant.
-
-## Risk Areas / Edge Cases
-- `tool_execution_update.partialResult` is cumulative; must compute deltas.
-- `message_update` may emit `done`/`error` without `message_end`; handle both paths.
-- No native session lifecycle events; rely on daemon synthetic events.
-- Session recovery after RPC server restart requires `get_state` + re-register sessions.
-
-## Acceptance Criteria
-- Pi appears in `/v1/agents`, CLI list, and docs.
-- `create_session` returns `native_session_id` from Pi `new_session`.
-- Streaming prompt yields universal events with proper ordering:
-  - message -> item.started/delta/completed
-  - tool execution -> tool call + tool result
-- Tests pass and no synthetic data is used in test fixtures.
-
-## Sources
-- https://upd.dev/badlogic/pi-mono/src/commit/d36e0ea07303d8a76d51b4a7bd5f0d6d3c490860/packages/coding-agent/docs/rpc.md
-- https://buildwithpi.ai/pi-cli
-- https://takopi.dev/docs/pi-cli/
-- https://upd.dev/badlogic/pi-mono/releases

docs/quickstart.mdx

@@ -64,7 +64,7 @@ icon: "rocket"
         docker run -p 2468:2468 \
           -e ANTHROPIC_API_KEY="sk-ant-..." \
           -e OPENAI_API_KEY="sk-..." \
-          rivetdev/sandbox-agent:0.3.1-full \
+          rivetdev/sandbox-agent:0.4.2-full \
           server --no-token --host 0.0.0.0 --port 2468
         ```
       </Tab>
@@ -77,6 +77,9 @@ icon: "rocket"
       <Accordion title="Testing without API keys">
         Use the `mock` agent for SDK and integration testing without provider credentials.
       </Accordion>
+      <Accordion title="Multi-tenant and per-user billing">
+        For per-tenant token tracking, budget enforcement, or usage-based billing, see [LLM Credentials](/llm-credentials) for gateway options like OpenRouter, LiteLLM, and Portkey.
+      </Accordion>
     </AccordionGroup>
   </Step>
 
@@ -86,7 +89,7 @@ icon: "rocket"
         Install and run the binary directly.
 
         ```bash
-        curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh
+        curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh
         sandbox-agent server --no-token --host 0.0.0.0 --port 2468
         ```
       </Tab>
@@ -95,7 +98,7 @@ icon: "rocket"
         Run without installing globally.
 
         ```bash
-        npx @sandbox-agent/cli@0.3.x server --no-token --host 0.0.0.0 --port 2468
+        npx @sandbox-agent/cli@0.4.x server --no-token --host 0.0.0.0 --port 2468
         ```
       </Tab>
 
@@ -103,7 +106,7 @@ icon: "rocket"
         Run without installing globally.
 
         ```bash
-        bunx @sandbox-agent/cli@0.3.x server --no-token --host 0.0.0.0 --port 2468
+        bunx @sandbox-agent/cli@0.4.x server --no-token --host 0.0.0.0 --port 2468
         ```
       </Tab>
 
@@ -111,7 +114,7 @@ icon: "rocket"
         Install globally, then run.
 
         ```bash
-        npm install -g @sandbox-agent/cli@0.3.x
+        npm install -g @sandbox-agent/cli@0.4.x
         sandbox-agent server --no-token --host 0.0.0.0 --port 2468
         ```
       </Tab>
@@ -120,7 +123,7 @@ icon: "rocket"
         Install globally, then run.
 
         ```bash
-        bun add -g @sandbox-agent/cli@0.3.x
+        bun add -g @sandbox-agent/cli@0.4.x
         # Allow Bun to run postinstall scripts for native binaries (required for SandboxAgent.start()).
         bun pm -g trust @sandbox-agent/cli-linux-x64 @sandbox-agent/cli-linux-arm64 @sandbox-agent/cli-darwin-arm64 @sandbox-agent/cli-darwin-x64 @sandbox-agent/cli-win32-x64
         sandbox-agent server --no-token --host 0.0.0.0 --port 2468
@@ -131,7 +134,7 @@ icon: "rocket"
         For local development, use `SandboxAgent.start()` to spawn and manage the server as a subprocess.
 
         ```bash
-        npm install sandbox-agent@0.3.x
+        npm install sandbox-agent@0.4.x
         ```
 
         ```typescript
@@ -145,7 +148,7 @@ icon: "rocket"
         For local development, use `SandboxAgent.start()` to spawn and manage the server as a subprocess.
 
         ```bash
-        bun add sandbox-agent@0.3.x
+        bun add sandbox-agent@0.4.x
         # Allow Bun to run postinstall scripts for native binaries (required for SandboxAgent.start()).
         bun pm trust @sandbox-agent/cli-linux-x64 @sandbox-agent/cli-linux-arm64 @sandbox-agent/cli-darwin-arm64 @sandbox-agent/cli-darwin-x64 @sandbox-agent/cli-win32-x64
         ```
@@ -223,6 +226,16 @@ icon: "rocket"
     If agents are not installed up front, they are lazily installed when creating a session.
   </Step>
 
+  <Step title="Install desktop dependencies (optional, Linux only)">
+    If you want to use `/v1/desktop/*`, install the desktop runtime packages first:
+
+    ```bash
+    sandbox-agent install desktop --yes
+    ```
+
+    Then use `GET /v1/desktop/status` or `sdk.getDesktopStatus()` to verify the runtime is ready before calling desktop screenshot or input APIs.
+  </Step>
+
   <Step title="Create a session">
     ```typescript
     import { SandboxAgent } from "sandbox-agent";

docs/react-components.mdx

@@ -12,11 +12,12 @@ Current exports:
 - `ProcessTerminal` for attaching to a running tty process
 - `AgentTranscript` for rendering session/message timelines without bundling any styles
 - `ChatComposer` for a reusable prompt input/send surface
+- `useTranscriptVirtualizer` for wiring large transcript lists to a scroll container
 
 ## Install
 
 ```bash
-npm install @sandbox-agent/react@0.3.x
+npm install @sandbox-agent/react@0.4.x
 ```
 
 ## Full example
@@ -184,11 +185,20 @@ Useful props:
 
 - `className`: root class hook
 - `classNames`: slot-level class hooks for styling from outside the package
+- `scrollRef` + `virtualize`: opt into TanStack Virtual against an external scroll container
 - `renderMessageText`: custom text or markdown renderer
 - `renderToolItemIcon`, `renderToolGroupIcon`, `renderChevron`, `renderEventLinkContent`: presentation overrides
 - `renderInlinePendingIndicator`, `renderThinkingState`: loading/thinking UI overrides
 - `isDividerEntry`, `canOpenEvent`, `getToolGroupSummary`: behavior overrides for grouping and labels
 
+## Transcript virtualization hook
+
+`useTranscriptVirtualizer` exposes the same TanStack Virtual behavior used by `AgentTranscript` when `virtualize` is enabled.
+
+- Pass the grouped transcript rows you want to virtualize
+- Pass a `scrollRef` that points at the actual scrollable element
+- Use it when you need transcript-aware virtualization outside the stock `AgentTranscript` renderer
+
 ## Composer and conversation
 
 `ChatComposer` is the headless message input. `AgentConversation` composes `AgentTranscript` and `ChatComposer` so apps can reuse the transcript/composer pairing without pulling in Inspector session chrome.

docs/sdk-overview.mdx

@@ -11,28 +11,22 @@ The TypeScript SDK is centered on `sandbox-agent` and its `SandboxAgent` class.
 <Tabs>
   <Tab title="npm">
     ```bash
-    npm install sandbox-agent@0.3.x
+    npm install sandbox-agent@0.4.x
     ```
   </Tab>
   <Tab title="bun">
     ```bash
-    bun add sandbox-agent@0.3.x
+    bun add sandbox-agent@0.4.x
     # Allow Bun to run postinstall scripts for native binaries (required for SandboxAgent.start()).
     bun pm trust @sandbox-agent/cli-linux-x64 @sandbox-agent/cli-linux-arm64 @sandbox-agent/cli-darwin-arm64 @sandbox-agent/cli-darwin-x64 @sandbox-agent/cli-win32-x64
     ```
   </Tab>
 </Tabs>
 
-## Optional persistence drivers
-
-```bash
-npm install @sandbox-agent/persist-indexeddb@0.3.x @sandbox-agent/persist-sqlite@0.3.x @sandbox-agent/persist-postgres@0.3.x
-```
-
 ## Optional React components
 
 ```bash
-npm install @sandbox-agent/react@0.3.x
+npm install @sandbox-agent/react@0.4.x
 ```
 
 ## Create a client
@@ -68,15 +62,12 @@ const sdk = await SandboxAgent.connect({
 controller.abort();
 ```
 
-With persistence:
+With persistence (see [Persisting Sessions](/session-persistence) for driver options):
 
 ```ts
-import { SandboxAgent } from "sandbox-agent";
+import { SandboxAgent, InMemorySessionPersistDriver } from "sandbox-agent";
-import { SQLiteSessionPersistDriver } from "@sandbox-agent/persist-sqlite";
 
-const persist = new SQLiteSessionPersistDriver({
+const persist = new InMemorySessionPersistDriver();
-  filename: "./sessions.db",
-});
 
 const sdk = await SandboxAgent.connect({
   baseUrl: "http://127.0.0.1:2468",
@@ -84,25 +75,40 @@ const sdk = await SandboxAgent.connect({
 });
 ```
 
-Local autospawn (Node.js only):
+Local spawn with a sandbox provider:
 
 ```ts
 import { SandboxAgent } from "sandbox-agent";
+import { local } from "sandbox-agent/local";
 
-const localSdk = await SandboxAgent.start();
+const sdk = await SandboxAgent.start({
+  sandbox: local(),
+});
 
-await localSdk.dispose();
+// sdk.sandboxId — prefixed provider ID (e.g. "local/127.0.0.1:2468")
+
+await sdk.destroySandbox(); // provider-defined cleanup + disposes client
 ```
 
+`SandboxAgent.start(...)` requires a `sandbox` provider. Built-in providers:
+
+| Import | Provider |
+|--------|----------|
+| `sandbox-agent/local` | Local subprocess |
+| `sandbox-agent/docker` | Docker container |
+| `sandbox-agent/e2b` | E2B sandbox |
+| `sandbox-agent/daytona` | Daytona workspace |
+| `sandbox-agent/vercel` | Vercel Sandbox |
+| `sandbox-agent/cloudflare` | Cloudflare Sandbox |
+
+Use `sdk.dispose()` to disconnect without changing sandbox state, `sdk.pauseSandbox()` for graceful suspension when supported, or `sdk.killSandbox()` for permanent deletion.
+
 ## Session flow
 
 ```ts
 const session = await sdk.createSession({
   agent: "mock",
-  sessionInit: {
+  cwd: "/",
-    cwd: "/",
-    mcpServers: [],
-  },
 });
 
 const prompt = await session.prompt([
@@ -190,6 +196,44 @@ const writeResult = await sdk.writeFsFile({ path: "./hello.txt" }, "hello");
 console.log(health.status, agents.agents.length, entries.length, writeResult.path);
 ```
 
+## Desktop API
+
+The SDK also wraps the desktop host/runtime HTTP API.
+
+Install desktop dependencies first on Linux hosts:
+
+```bash
+sandbox-agent install desktop --yes
+```
+
+Then query status, surface remediation if needed, and start the runtime:
+
+```ts
+const status = await sdk.getDesktopStatus();
+
+if (status.state === "install_required") {
+  console.log(status.installCommand);
+}
+
+const started = await sdk.startDesktop({
+  width: 1440,
+  height: 900,
+  dpi: 96,
+});
+
+const screenshot = await sdk.takeDesktopScreenshot();
+const displayInfo = await sdk.getDesktopDisplayInfo();
+
+await sdk.moveDesktopMouse({ x: 400, y: 300 });
+await sdk.clickDesktop({ x: 400, y: 300, button: "left", clickCount: 1 });
+await sdk.typeDesktopText({ text: "hello world", delayMs: 10 });
+await sdk.pressDesktopKey({ key: "ctrl+l" });
+
+await sdk.stopDesktop();
+```
+
+Screenshot helpers return `Uint8Array` PNG bytes. The SDK does not attempt to install OS packages remotely; callers should surface `missingDependencies` and `installCommand` from `getDesktopStatus()`.
+
 ## Error handling
 
 ```ts
@@ -223,5 +267,10 @@ Parameters:
 - `token` (optional): Bearer token for authenticated servers
 - `headers` (optional): Additional request headers
 - `fetch` (optional): Custom fetch implementation used by SDK HTTP and session calls
+- `skipHealthCheck` (optional): set `true` to skip the startup `/v1/health` wait
 - `waitForHealth` (optional, defaults to enabled): waits for `/v1/health` before HTTP helpers and session setup proceed; pass `false` to disable or `{ timeoutMs }` to bound the wait
 - `signal` (optional): aborts the startup `/v1/health` wait used by `connect()`
+
+## LLM credentials
+
+Sandbox Agent supports personal API keys, shared organization keys, and per-tenant gateway keys with budget enforcement. See [LLM Credentials](/llm-credentials) for setup details.

docs/security.mdx

@@ -4,7 +4,7 @@ description: "Backend-first auth and access control patterns."
 icon: "shield"
 ---
 
-As covered in [Architecture](/architecture), run the Sandbox Agent client on your backend, not in the browser.
+As covered in [Orchestration Architecture](/orchestration-architecture), run the Sandbox Agent client on your backend, not in the browser.
 
 This keeps sandbox credentials private and gives you one place for authz, rate limiting, and audit logging.
 
@@ -92,7 +92,7 @@ export const workspace = actor({
 
       const session = await sdk.createSession({
         agent: "claude",
-        sessionInit: { cwd: "/workspace" },
+        cwd: "/workspace",
       });
 
       session.onEvent((event) => {

docs/session-persistence.mdx

@@ -10,14 +10,22 @@ With persistence enabled, sessions can be restored after runtime/session loss. S
 
 Each driver stores:
 
-- `SessionRecord` (`id`, `agent`, `agentSessionId`, `lastConnectionId`, `createdAt`, optional `destroyedAt`, optional `sessionInit`)
+- `SessionRecord` (`id`, `agent`, `agentSessionId`, `lastConnectionId`, `createdAt`, optional `destroyedAt`, optional `sandboxId`, optional `sessionInit`, optional `configOptions`, optional `modes`)
 - `SessionEvent` (`id`, `eventIndex`, `sessionId`, `connectionId`, `sender`, `payload`, `createdAt`)
 
 ## Persistence drivers
 
-### In-memory
+### Rivet
 
-Best for local dev and ephemeral workloads.
+Recommended for sandbox orchestration with actor state. See [Multiplayer](/multiplayer) for a full Rivet actor example with persistence in actor state.
+
+### IndexedDB (browser)
+
+Best for browser apps that should survive reloads. See the [Inspector source](https://github.com/rivet-dev/sandbox-agent/tree/main/frontend/packages/inspector/src/persist-indexeddb.ts) for a complete IndexedDB driver you can copy into your project.
+
+### In-memory (built-in)
+
+Best for local dev and ephemeral workloads. No extra dependencies required.
 
 ```ts
 import { InMemorySessionPersistDriver, SandboxAgent } from "sandbox-agent";
@@ -33,91 +41,17 @@ const sdk = await SandboxAgent.connect({
 });
 ```
 
-### Rivet
-
-Recommended for sandbox orchestration with actor state.
-
-```bash
-npm install @sandbox-agent/persist-rivet@0.3.x
-```
-
-```ts
-import { actor } from "rivetkit";
-import { SandboxAgent } from "sandbox-agent";
-import { RivetSessionPersistDriver, type RivetPersistState } from "@sandbox-agent/persist-rivet";
-
-type PersistedState = RivetPersistState & {
-  sandboxId: string;
-  baseUrl: string;
-};
-
-export default actor({
-  createState: async () => {
-    return {
-      sandboxId: "sbx_123",
-      baseUrl: "http://127.0.0.1:2468",
-    } satisfies Partial<PersistedState>;
-  },
-  createVars: async (c) => {
-    const persist = new RivetSessionPersistDriver(c);
-    const sdk = await SandboxAgent.connect({
-      baseUrl: c.state.baseUrl,
-      persist,
-    });
-
-    const session = await sdk.resumeOrCreateSession({ id: "default", agent: "codex" });
-
-    const unsubscribe = session.onEvent((event) => {
-      c.broadcast("session.event", event);
-    });
-
-    return { sdk, session, unsubscribe };
-  },
-  actions: {
-    sendMessage: async (c, message: string) => {
-      await c.vars.session.prompt([{ type: "text", text: message }]);
-    },
-  },
-  onSleep: async (c) => {
-    c.vars.unsubscribe?.();
-    await c.vars.sdk.dispose();
-  },
-});
-```
-
-### IndexedDB
-
-Best for browser apps that should survive reloads.
-
-```bash
-npm install @sandbox-agent/persist-indexeddb@0.3.x
-```
-
-```ts
-import { SandboxAgent } from "sandbox-agent";
-import { IndexedDbSessionPersistDriver } from "@sandbox-agent/persist-indexeddb";
-
-const persist = new IndexedDbSessionPersistDriver({
-  databaseName: "sandbox-agent-session-store",
-});
-
-const sdk = await SandboxAgent.connect({
-  baseUrl: "http://127.0.0.1:2468",
-  persist,
-});
-```
-
 ### SQLite
 
 Best for local/server Node apps that need durable storage without a DB server.
 
 ```bash
-npm install @sandbox-agent/persist-sqlite@0.3.x
+npm install better-sqlite3
 ```
 
 ```ts
 import { SandboxAgent } from "sandbox-agent";
-import { SQLiteSessionPersistDriver } from "@sandbox-agent/persist-sqlite";
+import { SQLiteSessionPersistDriver } from "./persist.ts";
 
 const persist = new SQLiteSessionPersistDriver({
   filename: "./sandbox-agent.db",
@@ -129,17 +63,19 @@ const sdk = await SandboxAgent.connect({
 });
 ```
 
+See the [full SQLite example](https://github.com/rivet-dev/sandbox-agent/tree/main/examples/persist-sqlite) for the complete driver implementation you can copy into your project.
+
 ### Postgres
 
 Use when you already run Postgres and want shared relational storage.
 
 ```bash
-npm install @sandbox-agent/persist-postgres@0.3.x
+npm install pg
 ```
 
 ```ts
 import { SandboxAgent } from "sandbox-agent";
-import { PostgresSessionPersistDriver } from "@sandbox-agent/persist-postgres";
+import { PostgresSessionPersistDriver } from "./persist.ts";
 
 const persist = new PostgresSessionPersistDriver({
   connectionString: process.env.DATABASE_URL,
@@ -152,6 +88,8 @@ const sdk = await SandboxAgent.connect({
 });
 ```
 
+See the [full Postgres example](https://github.com/rivet-dev/sandbox-agent/tree/main/examples/persist-postgres) for the complete driver implementation you can copy into your project.
+
 ### Custom driver
 
 Implement `SessionPersistDriver` for custom backends.
@@ -160,11 +98,11 @@ Implement `SessionPersistDriver` for custom backends.
 import type { SessionPersistDriver } from "sandbox-agent";
 
 class MyDriver implements SessionPersistDriver {
-  async getSession(id) { return null; }
+  async getSession(id) { return undefined; }
   async listSessions(request) { return { items: [] }; }
   async updateSession(session) {}
   async listEvents(request) { return { items: [] }; }
-  async insertEvent(event) {}
+  async insertEvent(sessionId, event) {}
 }
 ```
 

docs/session-transcript-schema.mdx

@@ -1,388 +0,0 @@
----
-title: "Session Transcript Schema"
-description: "Universal event schema for session transcripts across all agents."
----
-
-Each coding agent outputs events in its own native format. The sandbox-agent converts these into a universal event schema, giving you a consistent session transcript regardless of which agent you use.
-
-The schema is defined in [OpenAPI format](https://github.com/rivet-dev/sandbox-agent/blob/main/docs/openapi.json). See the [HTTP API Reference](/api-reference) for endpoint documentation.
-
-## Coverage Matrix
-
-This table shows which agent feature coverage appears in the universal event stream. All agents retain their full native feature coverage—this only reflects what's normalized into the schema.
-
-| Feature            | Claude | Codex | OpenCode     | Amp          | Pi (RPC)     |
-|--------------------|:------:|:-----:|:------------:|:------------:|:------------:|
-| Stability          | Stable | Stable| Experimental | Experimental | Experimental |
-| Text Messages      |   ✓    |   ✓   |      ✓       |      ✓       |      ✓       |
-| Tool Calls         |   ✓    |   ✓   |      ✓       |      ✓       |      ✓       |
-| Tool Results       |   ✓    |   ✓   |      ✓       |      ✓       |      ✓       |
-| Questions (HITL)   |   ✓    |       |      ✓       |              |              |
-| Permissions (HITL) |   ✓    |   ✓   |      ✓       |      -       |              |
-| Images             |   -    |   ✓   |      ✓       |      -       |      ✓       |
-| File Attachments   |   -    |   ✓   |      ✓       |      -       |              |
-| Session Lifecycle  |   -    |   ✓   |      ✓       |      -       |              |
-| Error Events       |   -    |   ✓   |      ✓       |      ✓       |      ✓       |
-| Reasoning/Thinking |   -    |   ✓   |      -       |      -       |      ✓       |
-| Command Execution  |   -    |   ✓   |      -       |      -       |              |
-| File Changes       |   -    |   ✓   |      -       |      -       |              |
-| MCP Tools          |   ✓    |   ✓   |      ✓       |      ✓       |              |
-| Streaming Deltas   |   ✓    |   ✓   |      ✓       |      -       |      ✓       |
-| Variants           |        |   ✓   |      ✓       |      ✓       |      ✓       |
-
-Agents: [Claude Code](https://docs.anthropic.com/en/docs/agents-and-tools/claude-code/overview) · [Codex](https://github.com/openai/codex) · [OpenCode](https://github.com/opencode-ai/opencode) · [Amp](https://ampcode.com) · [Pi](https://buildwithpi.ai/pi-cli)
-
-- ✓ = Appears in session events
-- \- = Agent supports natively, schema conversion coming soon
-- (blank) = Not supported by agent
-- Pi runtime model is router-managed per-session RPC (`pi --mode rpc`); it does not use generic subprocess streaming.
-
-<AccordionGroup>
-  <Accordion title="Text Messages">
-    Basic message exchange between user and assistant.
-  </Accordion>
-  <Accordion title="Tool Calls & Results">
-    Visibility into tool invocations (file reads, command execution, etc.) and their results. When not natively supported, tool activity is embedded in message content.
-  </Accordion>
-  <Accordion title="Questions (HITL)">
-    Interactive questions the agent asks the user. Emits `question.requested` and `question.resolved` events.
-  </Accordion>
-  <Accordion title="Permissions (HITL)">
-    Permission requests for sensitive operations. Emits `permission.requested` and `permission.resolved` events.
-  </Accordion>
-  <Accordion title="Images">
-    Support for image attachments in messages.
-  </Accordion>
-  <Accordion title="File Attachments">
-    Support for file attachments in messages.
-  </Accordion>
-  <Accordion title="Session Lifecycle">
-    Native `session.started` and `session.ended` events. When not supported, the daemon emits synthetic lifecycle events.
-  </Accordion>
-  <Accordion title="Error Events">
-    Structured error events for runtime failures.
-  </Accordion>
-  <Accordion title="Reasoning/Thinking">
-    Extended thinking or reasoning content with visibility controls.
-  </Accordion>
-  <Accordion title="Command Execution">
-    Detailed command execution events with stdout/stderr.
-  </Accordion>
-  <Accordion title="File Changes">
-    Structured file modification events with diffs.
-  </Accordion>
-  <Accordion title="MCP Tools">
-    Model Context Protocol tool support.
-  </Accordion>
-  <Accordion title="Streaming Deltas">
-    Native streaming of content deltas. When not supported, the daemon emits a single synthetic delta before `item.completed`.
-  </Accordion>
-  <Accordion title="Variants">
-    Model variants such as reasoning effort or depth. Agents may expose different variant sets per model.
-  </Accordion>
-</AccordionGroup>
-
-Want support for another agent? [Open an issue](https://github.com/rivet-dev/sandbox-agent/issues/new) to request it.
-
-## UniversalEvent
-
-Every event from the API is wrapped in a `UniversalEvent` envelope.
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `event_id` | string | Unique identifier for this event |
-| `sequence` | integer | Monotonic sequence number within the session (starts at 1) |
-| `time` | string | RFC3339 timestamp |
-| `session_id` | string | Daemon-generated session identifier |
-| `native_session_id` | string? | Provider-native session/thread identifier (e.g., Codex `threadId`, OpenCode `sessionID`) |
-| `source` | string | Event origin: `agent` (native) or `daemon` (synthetic) |
-| `synthetic` | boolean | Whether this event was generated by the daemon to fill gaps |
-| `type` | string | Event type (see [Event Types](#event-types)) |
-| `data` | object | Event-specific payload |
-| `raw` | any? | Original provider payload (only when `include_raw=true`) |
-
-```json
-{
-  "event_id": "evt_abc123",
-  "sequence": 1,
-  "time": "2025-01-28T12:00:00Z",
-  "session_id": "my-session",
-  "native_session_id": "thread_xyz",
-  "source": "agent",
-  "synthetic": false,
-  "type": "item.completed",
-  "data": { ... }
-}
-```
-
-## Event Types
-
-### Session Lifecycle
-
-| Type | Description | Data |
-|------|-------------|------|
-| `session.started` | Session has started | `{ metadata?: any }` |
-| `session.ended` | Session has ended | `{ reason, terminated_by, message?, exit_code? }` |
-
-### Turn Lifecycle
-
-| Type | Description | Data |
-|------|-------------|------|
-| `turn.started` | Turn has started | `{ phase: "started", turn_id?, metadata? }` |
-| `turn.ended` | Turn has ended | `{ phase: "ended", turn_id?, metadata? }` |
-
-**SessionEndedData**
-
-| Field | Type | Values |
-|-------|------|--------|
-| `reason` | string | `completed`, `error`, `terminated` |
-| `terminated_by` | string | `agent`, `daemon` |
-| `message` | string? | Error message (only present when reason is `error`) |
-| `exit_code` | int? | Process exit code (only present when reason is `error`) |
-| `stderr` | StderrOutput? | Structured stderr output (only present when reason is `error`) |
-
-**StderrOutput**
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `head` | string? | First 20 lines of stderr (if truncated) or full stderr (if not truncated) |
-| `tail` | string? | Last 50 lines of stderr (only present if truncated) |
-| `truncated` | boolean | Whether the output was truncated |
-| `total_lines` | int? | Total number of lines in stderr |
-
-### Item Lifecycle
-
-| Type | Description | Data |
-|------|-------------|------|
-| `item.started` | Item creation | `{ item }` |
-| `item.delta` | Streaming content delta | `{ item_id, native_item_id?, delta }` |
-| `item.completed` | Item finalized | `{ item }` |
-
-Items follow a consistent lifecycle: `item.started` → `item.delta` (0 or more) → `item.completed`.
-
-### HITL (Human-in-the-Loop)
-
-| Type | Description | Data |
-|------|-------------|------|
-| `permission.requested` | Permission request pending | `{ permission_id, action, status, metadata? }` |
-| `permission.resolved` | Permission decision recorded | `{ permission_id, action, status, metadata? }` |
-| `question.requested` | Question pending user input | `{ question_id, prompt, options, status }` |
-| `question.resolved` | Question answered or rejected | `{ question_id, prompt, options, status, response? }` |
-
-**PermissionEventData**
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `permission_id` | string | Identifier for the permission request |
-| `action` | string | What the agent wants to do |
-| `status` | string | `requested`, `accept`, `accept_for_session`, `reject` |
-| `metadata` | any? | Additional context |
-
-**QuestionEventData**
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `question_id` | string | Identifier for the question |
-| `prompt` | string | Question text |
-| `options` | string[] | Available answer options |
-| `status` | string | `requested`, `answered`, `rejected` |
-| `response` | string? | Selected answer (when resolved) |
-
-### Errors
-
-| Type | Description | Data |
-|------|-------------|------|
-| `error` | Runtime error | `{ message, code?, details? }` |
-| `agent.unparsed` | Parse failure | `{ error, location, raw_hash? }` |
-
-The `agent.unparsed` event indicates the daemon failed to parse an agent payload. This should be treated as a bug.
-
-## UniversalItem
-
-Items represent discrete units of content within a session.
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `item_id` | string | Daemon-generated identifier |
-| `native_item_id` | string? | Provider-native item/message identifier |
-| `parent_id` | string? | Parent item ID (e.g., tool call/result parented to a message) |
-| `kind` | string | Item category (see below) |
-| `role` | string? | Actor role for message items |
-| `status` | string | Lifecycle status |
-| `content` | ContentPart[] | Ordered list of content parts |
-
-### ItemKind
-
-| Value | Description |
-|-------|-------------|
-| `message` | User or assistant message |
-| `tool_call` | Tool invocation |
-| `tool_result` | Tool execution result |
-| `system` | System message |
-| `status` | Status update |
-| `unknown` | Unrecognized item type |
-
-### ItemRole
-
-| Value | Description |
-|-------|-------------|
-| `user` | User message |
-| `assistant` | Assistant response |
-| `system` | System prompt |
-| `tool` | Tool-related message |
-
-### ItemStatus
-
-| Value | Description |
-|-------|-------------|
-| `in_progress` | Item is streaming or pending |
-| `completed` | Item is finalized |
-| `failed` | Item execution failed |
-
-## Content Parts
-
-The `content` array contains typed parts that make up an item's payload.
-
-### text
-
-Plain text content.
-
-```json
-{ "type": "text", "text": "Hello, world!" }
-```
-
-### json
-
-Structured JSON content.
-
-```json
-{ "type": "json", "json": { "key": "value" } }
-```
-
-### tool_call
-
-Tool invocation.
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `name` | string | Tool name |
-| `arguments` | string | JSON-encoded arguments |
-| `call_id` | string | Unique call identifier |
-
-```json
-{
-  "type": "tool_call",
-  "name": "read_file",
-  "arguments": "{\"path\": \"/src/main.ts\"}",
-  "call_id": "call_abc123"
-}
-```
-
-### tool_result
-
-Tool execution result.
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `call_id` | string | Matching call identifier |
-| `output` | string | Tool output |
-
-```json
-{
-  "type": "tool_result",
-  "call_id": "call_abc123",
-  "output": "File contents here..."
-}
-```
-
-### file_ref
-
-File reference with optional diff.
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `path` | string | File path |
-| `action` | string | `read`, `write`, `patch` |
-| `diff` | string? | Unified diff (for patches) |
-
-```json
-{
-  "type": "file_ref",
-  "path": "/src/main.ts",
-  "action": "write",
-  "diff": "@@ -1,3 +1,4 @@\n+import { foo } from 'bar';"
-}
-```
-
-### image
-
-Image reference.
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `path` | string | Image file path |
-| `mime` | string? | MIME type |
-
-```json
-{ "type": "image", "path": "/tmp/screenshot.png", "mime": "image/png" }
-```
-
-### reasoning
-
-Model reasoning/thinking content.
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `text` | string | Reasoning text |
-| `visibility` | string | `public` or `private` |
-
-```json
-{ "type": "reasoning", "text": "Let me think about this...", "visibility": "public" }
-```
-
-### status
-
-Status indicator.
-
-| Field | Type | Description |
-|-------|------|-------------|
-| `label` | string | Status label |
-| `detail` | string? | Additional detail |
-
-```json
-{ "type": "status", "label": "Running tests", "detail": "3 of 10 passed" }
-```
-
-## Source & Synthetics
-
-### EventSource
-
-The `source` field indicates who emitted the event:
-
-| Value | Description |
-|-------|-------------|
-| `agent` | Native event from the agent |
-| `daemon` | Synthetic event generated by the daemon |
-
-### Synthetic Events
-
-The daemon emits synthetic events (`synthetic: true`, `source: "daemon"`) to provide a consistent event stream across all agents. Common synthetics:
-
-| Synthetic | When |
-|-----------|------|
-| `session.started` | Agent doesn't emit explicit session start |
-| `session.ended` | Agent doesn't emit explicit session end |
-| `turn.started` | Agent doesn't emit explicit turn start |
-| `turn.ended` | Agent doesn't emit explicit turn end |
-| `item.started` | Agent doesn't emit item start events |
-| `item.delta` | Agent doesn't stream deltas natively |
-| `question.*` | Claude Code plan mode (from ExitPlanMode tool) |
-
-### Raw Payloads
-
-Pass `include_raw=true` to event endpoints to receive the original agent payload in the `raw` field. Useful for debugging or accessing agent-specific data not in the universal schema.
-
-```typescript
-const events = await client.getEvents("my-session", { includeRaw: true });
-// events[0].raw contains the original agent payload
-```

docs/skills-config.mdx

@@ -35,9 +35,7 @@ await sdk.setSkillsConfig(
 // Create a session using the configured skills
 const session = await sdk.createSession({
   agent: "claude",
-  sessionInit: {
+  cwd: "/workspace",
-    cwd: "/workspace",
-  },
 });
 
 await session.prompt([

docs/theme.css

@@ -20,7 +20,6 @@ body {
   color: var(--sa-text);
 }
 
-/*
 a {
   color: var(--sa-primary);
 }
@@ -41,6 +40,13 @@ select {
   color: var(--sa-text);
 }
 
+code,
+pre {
+  background-color: var(--sa-card);
+  border: 1px solid var(--sa-border);
+  color: var(--sa-text);
+}
+
 .card,
 .mintlify-card,
 .docs-card {
@@ -64,4 +70,3 @@ select {
 .alert-danger {
   border-color: var(--sa-danger);
 }
-*/

docs/troubleshooting.mdx

@@ -29,25 +29,6 @@ Verify the agent is installed:
 ls -la ~/.local/share/sandbox-agent/bin/
 ```
 
-### 4. Binary libc mismatch (musl vs glibc)
-
-Claude Code binaries are available in both musl and glibc variants. If you see errors like:
-
-```
-cannot execute: required file not found
-Error loading shared library libstdc++.so.6: No such file or directory
-```
-
-This means the wrong binary variant was downloaded.
-
-**For sandbox-agent 0.2.0+**: Platform detection is automatic. The correct binary (musl or glibc) is downloaded based on the runtime environment.
-
-**For sandbox-agent 0.1.x**: Use Alpine Linux which has native musl support:
-
-```dockerfile
-FROM alpine:latest
-RUN apk add --no-cache curl ca-certificates libstdc++ libgcc bash
-```
 
 ## Daytona Network Restrictions
 

examples/boxlite-python/Dockerfile

@@ -1,5 +1,5 @@
 FROM node:22-bookworm-slim
 RUN apt-get update && apt-get install -y curl ca-certificates && rm -rf /var/lib/apt/lists/*
-RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh
+RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh
 RUN sandbox-agent install-agent claude
 RUN sandbox-agent install-agent codex

examples/boxlite/Dockerfile

@@ -1,5 +1,5 @@
 FROM node:22-bookworm-slim
 RUN apt-get update && apt-get install -y curl ca-certificates && rm -rf /var/lib/apt/lists/*
-RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh
+RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh
 RUN sandbox-agent install-agent claude
 RUN sandbox-agent install-agent codex

examples/boxlite/src/index.ts

@@ -25,7 +25,7 @@ const baseUrl = "http://localhost:3000";
 
 console.log("Connecting to server...");
 const client = await SandboxAgent.connect({ baseUrl });
-const session = await client.createSession({ agent: detectAgent(), sessionInit: { cwd: "/root", mcpServers: [] } });
+const session = await client.createSession({ agent: detectAgent(), cwd: "/root" });
 const sessionId = session.id;
 
 console.log(`  UI: ${buildInspectorUrl({ baseUrl, sessionId })}`);

examples/boxlite/tsconfig.json

@@ -9,7 +9,8 @@
     "esModuleInterop": true,
     "strict": true,
     "skipLibCheck": true,
-    "resolveJsonModule": true
+    "resolveJsonModule": true,
+    "types": ["node"]
   },
   "include": ["src/**/*"],
   "exclude": ["node_modules", "**/*.test.ts"]

examples/cloudflare/Dockerfile

@@ -1,7 +1,7 @@
 FROM cloudflare/sandbox:0.7.0
 
 # Install sandbox-agent
-RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh
+RUN curl -fsSL https://releases.rivet.dev/sandbox-agent/0.4.x/install.sh | sh
 
 # Pre-install agents
 RUN sandbox-agent install-agent claude && \

examples/cloudflare/tests/cloudflare.test.ts

@@ -0,0 +1,154 @@
+import { describe, it, expect } from "vitest";
+import { spawn, type ChildProcess } from "node:child_process";
+import { resolve, dirname } from "node:path";
+import { fileURLToPath } from "node:url";
+import { execSync } from "node:child_process";
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+const PROJECT_DIR = resolve(__dirname, "..");
+
+/**
+ * Cloudflare Workers integration test.
+ *
+ * Set RUN_CLOUDFLARE_EXAMPLES=1 to enable. Requires wrangler and Docker.
+ *
+ * This starts `wrangler dev` which:
+ *  1. Builds the Dockerfile (cloudflare/sandbox base + sandbox-agent)
+ *  2. Starts a local Workers runtime with Durable Objects and containers
+ *  3. Exposes the app on a local port
+ *
+ * We then test through the proxy endpoint which forwards to sandbox-agent
+ * running inside the container.
+ */
+const shouldRun = process.env.RUN_CLOUDFLARE_EXAMPLES === "1";
+const timeoutMs = Number.parseInt(process.env.SANDBOX_TEST_TIMEOUT_MS || "", 10) || 600_000;
+
+const testFn = shouldRun ? it : it.skip;
+
+interface WranglerDev {
+  baseUrl: string;
+  cleanup: () => void;
+}
+
+async function startWranglerDev(): Promise<WranglerDev> {
+  // Build frontend assets first (wrangler expects dist/ to exist)
+  execSync("npx vite build", { cwd: PROJECT_DIR, stdio: "pipe" });
+
+  return new Promise<WranglerDev>((resolve, reject) => {
+    const child: ChildProcess = spawn("npx", ["wrangler", "dev", "--port", "0"], {
+      cwd: PROJECT_DIR,
+      stdio: ["ignore", "pipe", "pipe"],
+      detached: true,
+      env: {
+        ...process.env,
+        // Ensure wrangler picks up API keys to pass to the container
+        NODE_ENV: "development",
+      },
+    });
+
+    let stdout = "";
+    let stderr = "";
+    let resolved = false;
+
+    const cleanup = () => {
+      if (child.pid) {
+        // Kill process group to ensure wrangler and its children are cleaned up
+        try {
+          process.kill(-child.pid, "SIGTERM");
+        } catch {
+          try {
+            child.kill("SIGTERM");
+          } catch {}
+        }
+      }
+    };
+
+    const timer = setTimeout(() => {
+      if (!resolved) {
+        resolved = true;
+        cleanup();
+        reject(new Error(`wrangler dev did not start within 120s.\nstdout: ${stdout}\nstderr: ${stderr}`));
+      }
+    }, 120_000);
+
+    const onData = (chunk: Buffer) => {
+      const text = chunk.toString();
+      stdout += text;
+
+      // wrangler dev prints "Ready on http://localhost:XXXX" when ready
+      const match = stdout.match(/Ready on (https?:\/\/[^\s]+)/i) ?? stdout.match(/(https?:\/\/(?:localhost|127\.0\.0\.1):\d+)/);
+      if (match && !resolved) {
+        resolved = true;
+        clearTimeout(timer);
+        resolve({ baseUrl: match[1], cleanup });
+      }
+    };
+
+    child.stdout?.on("data", onData);
+    child.stderr?.on("data", (chunk: Buffer) => {
+      const text = chunk.toString();
+      stderr += text;
+      // Some wrangler versions print ready message to stderr
+      const match = text.match(/Ready on (https?:\/\/[^\s]+)/i) ?? text.match(/(https?:\/\/(?:localhost|127\.0\.0\.1):\d+)/);
+      if (match && !resolved) {
+        resolved = true;
+        clearTimeout(timer);
+        resolve({ baseUrl: match[1], cleanup });
+      }
+    });
+
+    child.on("error", (err) => {
+      if (!resolved) {
+        resolved = true;
+        clearTimeout(timer);
+        reject(new Error(`wrangler dev failed to start: ${err.message}`));
+      }
+    });
+
+    child.on("exit", (code) => {
+      if (!resolved) {
+        resolved = true;
+        clearTimeout(timer);
+        reject(new Error(`wrangler dev exited with code ${code}.\nstdout: ${stdout}\nstderr: ${stderr}`));
+      }
+    });
+  });
+}
+
+describe("cloudflare example", () => {
+  testFn(
+    "starts wrangler dev and sandbox-agent responds via proxy",
+    async () => {
+      const { baseUrl, cleanup } = await startWranglerDev();
+      try {
+        // The Cloudflare example proxies requests through /sandbox/:name/proxy/*
+        // Wait for the container inside the Durable Object to start sandbox-agent
+        const healthUrl = `${baseUrl}/sandbox/test/proxy/v1/health`;
+
+        let healthy = false;
+        for (let i = 0; i < 120; i++) {
+          try {
+            const res = await fetch(healthUrl);
+            if (res.ok) {
+              const data = await res.json();
+              // The proxied health endpoint returns {name: "Sandbox Agent", ...}
+              if (data.status === "ok" || data.name === "Sandbox Agent") {
+                healthy = true;
+                break;
+              }
+            }
+          } catch {}
+          await new Promise((r) => setTimeout(r, 2000));
+        }
+        expect(healthy).toBe(true);
+
+        // Confirm a second request also works
+        const response = await fetch(healthUrl);
+        expect(response.ok).toBe(true);
+      } finally {
+        cleanup();
+      }
+    },
+    timeoutMs,
+  );
+});

examples/cloudflare/vitest.config.ts

@@ -2,7 +2,7 @@ import { defineConfig } from "vitest/config";
 
 export default defineConfig({
   test: {
+    root: ".",
     include: ["tests/**/*.test.ts"],
-    testTimeout: 60000,
   },
 });

examples/computesdk/package.json

@@ -3,7 +3,7 @@
   "private": true,
   "type": "module",
   "scripts": {
-    "start": "tsx src/computesdk.ts",
+    "start": "tsx src/index.ts",
     "typecheck": "tsc --noEmit"
   },
   "dependencies": {

examples/computesdk/src/computesdk.ts

@@ -1,151 +0,0 @@
-import {
-  compute,
-  detectProvider,
-  getMissingEnvVars,
-  getProviderConfigFromEnv,
-  isProviderAuthComplete,
-  isValidProvider,
-  PROVIDER_NAMES,
-  type ExplicitComputeConfig,
-  type ProviderName,
-} from "computesdk";
-import { SandboxAgent } from "sandbox-agent";
-import { detectAgent, buildInspectorUrl } from "@sandbox-agent/example-shared";
-import { fileURLToPath } from "node:url";
-import { resolve } from "node:path";
-
-const PORT = 3000;
-const REQUEST_TIMEOUT_MS = Number.parseInt(process.env.COMPUTESDK_TIMEOUT_MS || "", 10) || 120_000;
-
-/**
- * Detects and validates the provider to use.
- * Priority: COMPUTESDK_PROVIDER env var > auto-detection from API keys
- */
-function resolveProvider(): ProviderName {
-  const providerOverride = process.env.COMPUTESDK_PROVIDER;
-
-  if (providerOverride) {
-    if (!isValidProvider(providerOverride)) {
-      throw new Error(`Unsupported ComputeSDK provider "${providerOverride}". Supported providers: ${PROVIDER_NAMES.join(", ")}`);
-    }
-    if (!isProviderAuthComplete(providerOverride)) {
-      const missing = getMissingEnvVars(providerOverride);
-      throw new Error(`Missing credentials for provider "${providerOverride}". Set: ${missing.join(", ")}`);
-    }
-    console.log(`Using ComputeSDK provider: ${providerOverride} (explicit)`);
-    return providerOverride as ProviderName;
-  }
-
-  const detected = detectProvider();
-  if (!detected) {
-    throw new Error(`No provider credentials found. Set one of: ${PROVIDER_NAMES.map((p) => getMissingEnvVars(p).join(", ")).join(" | ")}`);
-  }
-  console.log(`Using ComputeSDK provider: ${detected} (auto-detected)`);
-  return detected as ProviderName;
-}
-
-function configureComputeSDK(): void {
-  const provider = resolveProvider();
-
-  const config: ExplicitComputeConfig = {
-    provider,
-    computesdkApiKey: process.env.COMPUTESDK_API_KEY,
-    requestTimeoutMs: REQUEST_TIMEOUT_MS,
-  };
-
-  const providerConfig = getProviderConfigFromEnv(provider);
-  if (Object.keys(providerConfig).length > 0) {
-    const configWithProvider = config as ExplicitComputeConfig & Record<ProviderName, Record<string, string>>;
-    configWithProvider[provider] = providerConfig;
-  }
-
-  compute.setConfig(config);
-}
-
-configureComputeSDK();
-
-const buildEnv = (): Record<string, string> => {
-  const env: Record<string, string> = {};
-  if (process.env.ANTHROPIC_API_KEY) env.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
-  if (process.env.OPENAI_API_KEY) env.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
-  return env;
-};
-
-export async function setupComputeSdkSandboxAgent(): Promise<{
-  baseUrl: string;
-  cleanup: () => Promise<void>;
-}> {
-  const env = buildEnv();
-
-  console.log("Creating ComputeSDK sandbox...");
-  const sandbox = await compute.sandbox.create({
-    envs: Object.keys(env).length > 0 ? env : undefined,
-  });
-
-  const run = async (cmd: string, options?: { background?: boolean }) => {
-    const result = await sandbox.runCommand(cmd, options);
-    if (typeof result?.exitCode === "number" && result.exitCode !== 0) {
-      throw new Error(`Command failed: ${cmd} (exit ${result.exitCode})\n${result.stderr || ""}`);
-    }
-    return result;
-  };
-
-  console.log("Installing sandbox-agent...");
-  await run("curl -fsSL https://releases.rivet.dev/sandbox-agent/latest/install.sh | sh");
-
-  if (env.ANTHROPIC_API_KEY) {
-    console.log("Installing Claude agent...");
-    await run("sandbox-agent install-agent claude");
-  }
-
-  if (env.OPENAI_API_KEY) {
-    console.log("Installing Codex agent...");
-    await run("sandbox-agent install-agent codex");
-  }
-
-  console.log("Starting server...");
-  await run(`sandbox-agent server --no-token --host 0.0.0.0 --port ${PORT}`, { background: true });
-
-  const baseUrl = await sandbox.getUrl({ port: PORT });
-
-  const cleanup = async () => {
-    try {
-      await sandbox.destroy();
-    } catch (error) {
-      console.warn("Cleanup failed:", error instanceof Error ? error.message : error);
-    }
-  };
-
-  return { baseUrl, cleanup };
-}
-
-export async function runComputeSdkExample(): Promise<void> {
-  const { baseUrl, cleanup } = await setupComputeSdkSandboxAgent();
-
-  const handleExit = async () => {
-    await cleanup();
-    process.exit(0);
-  };
-
-  process.once("SIGINT", handleExit);
-  process.once("SIGTERM", handleExit);
-
-  const client = await SandboxAgent.connect({ baseUrl });
-  const session = await client.createSession({ agent: detectAgent(), sessionInit: { cwd: "/home", mcpServers: [] } });
-  const sessionId = session.id;
-
-  console.log(`  UI: ${buildInspectorUrl({ baseUrl, sessionId })}`);
-  console.log("  Press Ctrl+C to stop.");
-
-  // Keep alive until SIGINT/SIGTERM triggers cleanup above
-  await new Promise(() => {});
-}
-
-const isDirectRun = Boolean(process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.url));
-
-if (isDirectRun) {
-  runComputeSdkExample().catch((error) => {
-    console.error(error instanceof Error ? error.message : error);
-    process.exit(1);
-  });
-}

examples/computesdk/src/index.ts

@@ -0,0 +1,30 @@
+import { SandboxAgent } from "sandbox-agent";
+import { computesdk } from "sandbox-agent/computesdk";
+import { detectAgent } from "@sandbox-agent/example-shared";
+
+const envs: Record<string, string> = {};
+if (process.env.ANTHROPIC_API_KEY) envs.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
+if (process.env.OPENAI_API_KEY) envs.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+
+const client = await SandboxAgent.start({
+  sandbox: computesdk({
+    create: { envs },
+  }),
+});
+
+console.log(`UI: ${client.inspectorUrl}`);
+
+const session = await client.createSession({
+  agent: detectAgent(),
+});
+
+session.onEvent((event) => {
+  console.log(`[${event.sender}]`, JSON.stringify(event.payload));
+});
+
+session.prompt([{ type: "text", text: "Say hello from ComputeSDK in one sentence." }]);
+
+process.once("SIGINT", async () => {
+  await client.destroySandbox();
+  process.exit(0);
+});

examples/computesdk/tests/computesdk.test.ts

@@ -1,6 +1,6 @@
 import { describe, it, expect } from "vitest";
-import { buildHeaders } from "@sandbox-agent/example-shared";
+import { SandboxAgent } from "sandbox-agent";
-import { setupComputeSdkSandboxAgent } from "../src/computesdk.ts";
+import { computesdk } from "sandbox-agent/computesdk";
 
 const hasModal = Boolean(process.env.MODAL_TOKEN_ID && process.env.MODAL_TOKEN_SECRET);
 const hasVercel = Boolean(process.env.VERCEL_TOKEN || process.env.VERCEL_OIDC_TOKEN);
@@ -13,20 +13,23 @@ const timeoutMs = Number.parseInt(process.env.SANDBOX_TEST_TIMEOUT_MS || "", 10)
 
 const testFn = shouldRun ? it : it.skip;
 
-describe("computesdk example", () => {
+describe("computesdk provider", () => {
   testFn(
     "starts sandbox-agent and responds to /v1/health",
     async () => {
-      const { baseUrl, cleanup } = await setupComputeSdkSandboxAgent();
+      const envs: Record<string, string> = {};
+      if (process.env.ANTHROPIC_API_KEY) envs.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
+      if (process.env.OPENAI_API_KEY) envs.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+
+      const sdk = await SandboxAgent.start({
+        sandbox: computesdk({ create: { envs } }),
+      });
+
       try {
-        const response = await fetch(`${baseUrl}/v1/health`, {
+        const health = await sdk.getHealth();
-          headers: buildHeaders({}),
+        expect(health.status).toBe("ok");
-        });
-        expect(response.ok).toBe(true);
-        const data = await response.json();
-        expect(data.status).toBe("ok");
       } finally {
-        await cleanup();
+        await sdk.destroySandbox();
       }
     },
     timeoutMs,

examples/computesdk/tsconfig.json

@@ -9,7 +9,8 @@
     "esModuleInterop": true,
     "strict": true,
     "skipLibCheck": true,
-    "resolveJsonModule": true
+    "resolveJsonModule": true,
+    "types": ["node"]
   },
   "include": ["src/**/*"],
   "exclude": ["node_modules", "**/*.test.ts"]

examples/daytona/src/daytona.ts

@@ -0,0 +1,33 @@
+import { SandboxAgent } from "sandbox-agent";
+import { daytona } from "sandbox-agent/daytona";
+
+function collectEnvVars(): Record<string, string> {
+  const envVars: Record<string, string> = {};
+  if (process.env.ANTHROPIC_API_KEY) envVars.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
+  if (process.env.OPENAI_API_KEY) envVars.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+  return envVars;
+}
+
+function inspectorUrlToBaseUrl(inspectorUrl: string): string {
+  return inspectorUrl.replace(/\/ui\/$/, "");
+}
+
+export async function setupDaytonaSandboxAgent(): Promise<{
+  baseUrl: string;
+  token?: string;
+  extraHeaders?: Record<string, string>;
+  cleanup: () => Promise<void>;
+}> {
+  const client = await SandboxAgent.start({
+    sandbox: daytona({
+      create: { envVars: collectEnvVars() },
+    }),
+  });
+
+  return {
+    baseUrl: inspectorUrlToBaseUrl(client.inspectorUrl),
+    cleanup: async () => {
+      await client.killSandbox();
+    },
+  };
+}

examples/daytona/src/index.ts

@@ -1,42 +1,30 @@
-import { Daytona } from "@daytonaio/sdk";
 import { SandboxAgent } from "sandbox-agent";
-import { detectAgent, buildInspectorUrl } from "@sandbox-agent/example-shared";
+import { daytona } from "sandbox-agent/daytona";
-
+import { detectAgent } from "@sandbox-agent/example-shared";
-const daytona = new Daytona();
 
 const envVars: Record<string, string> = {};
 if (process.env.ANTHROPIC_API_KEY) envVars.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
 if (process.env.OPENAI_API_KEY) envVars.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
 
-// Use default image and install sandbox-agent at runtime (faster startup, no snapshot build)
+const client = await SandboxAgent.start({
-console.log("Creating Daytona sandbox...");
+  sandbox: daytona({
-const sandbox = await daytona.create({ envVars, autoStopInterval: 0 });
+    create: { envVars },
+  }),
+});
 
-// Install sandbox-agent and start server
+console.log(`UI: ${client.inspectorUrl}`);
-console.log("Installing sandbox-agent...");
-await sandbox.process.executeCommand("curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh");
 
-console.log("Installing agents...");
+const session = await client.createSession({
-await sandbox.process.executeCommand("sandbox-agent install-agent claude");
+  agent: detectAgent(),
-await sandbox.process.executeCommand("sandbox-agent install-agent codex");
+});
 
-await sandbox.process.executeCommand("nohup sandbox-agent server --no-token --host 0.0.0.0 --port 3000 >/tmp/sandbox-agent.log 2>&1 &");
+session.onEvent((event) => {
+  console.log(`[${event.sender}]`, JSON.stringify(event.payload));
+});
 
-const baseUrl = (await sandbox.getSignedPreviewUrl(3000, 4 * 60 * 60)).url;
+session.prompt([{ type: "text", text: "Say hello from Daytona in one sentence." }]);
 
-console.log("Connecting to server...");
+process.once("SIGINT", async () => {
-const client = await SandboxAgent.connect({ baseUrl });
+  await client.destroySandbox();
-const session = await client.createSession({ agent: detectAgent(), sessionInit: { cwd: "/home/daytona", mcpServers: [] } });
-const sessionId = session.id;
-
-console.log(`  UI: ${buildInspectorUrl({ baseUrl, sessionId })}`);
-console.log("  Press Ctrl+C to stop.");
-
-const keepAlive = setInterval(() => {}, 60_000);
-const cleanup = async () => {
-  clearInterval(keepAlive);
-  await sandbox.delete(60);
   process.exit(0);
-};
+});
-process.once("SIGINT", cleanup);
-process.once("SIGTERM", cleanup);

examples/daytona/tsconfig.json

@@ -9,7 +9,8 @@
     "esModuleInterop": true,
     "strict": true,
     "skipLibCheck": true,
-    "resolveJsonModule": true
+    "resolveJsonModule": true,
+    "types": ["node"]
   },
   "include": ["src/**/*"],
   "exclude": ["node_modules", "**/*.test.ts"]

examples/docker/package.json

@@ -9,10 +9,10 @@
   "dependencies": {
     "@sandbox-agent/example-shared": "workspace:*",
     "dockerode": "latest",
+    "get-port": "latest",
     "sandbox-agent": "workspace:*"
   },
   "devDependencies": {
-    "@types/dockerode": "latest",
     "@types/node": "latest",
     "tsx": "latest",
     "typescript": "latest",

examples/docker/src/index.ts

@@ -1,68 +1,40 @@
-import Docker from "dockerode";
 import fs from "node:fs";
 import path from "node:path";
 import { SandboxAgent } from "sandbox-agent";
-import { detectAgent, buildInspectorUrl } from "@sandbox-agent/example-shared";
+import { docker } from "sandbox-agent/docker";
+import { detectAgent } from "@sandbox-agent/example-shared";
 import { FULL_IMAGE } from "@sandbox-agent/example-shared/docker";
 
-const IMAGE = FULL_IMAGE;
-const PORT = 3000;
-const agent = detectAgent();
 const codexAuthPath = process.env.HOME ? path.join(process.env.HOME, ".codex", "auth.json") : null;
 const bindMounts = codexAuthPath && fs.existsSync(codexAuthPath) ? [`${codexAuthPath}:/home/sandbox/.codex/auth.json:ro`] : [];
+const env = [
+  process.env.ANTHROPIC_API_KEY ? `ANTHROPIC_API_KEY=${process.env.ANTHROPIC_API_KEY}` : "",
+  process.env.OPENAI_API_KEY ? `OPENAI_API_KEY=${process.env.OPENAI_API_KEY}` : "",
+  process.env.CODEX_API_KEY ? `CODEX_API_KEY=${process.env.CODEX_API_KEY}` : "",
+].filter(Boolean);
 
-const docker = new Docker({ socketPath: "/var/run/docker.sock" });
+const client = await SandboxAgent.start({
-
+  sandbox: docker({
-// Pull image if needed
+    image: FULL_IMAGE,
-try {
+    env,
-  await docker.getImage(IMAGE).inspect();
+    binds: bindMounts,
-} catch {
+  }),
-  console.log(`Pulling ${IMAGE}...`);
-  await new Promise<void>((resolve, reject) => {
-    docker.pull(IMAGE, (err: Error | null, stream: NodeJS.ReadableStream) => {
-      if (err) return reject(err);
-      docker.modem.followProgress(stream, (err: Error | null) => (err ? reject(err) : resolve()));
-    });
-  });
-}
-
-console.log("Starting container...");
-const container = await docker.createContainer({
-  Image: IMAGE,
-  Cmd: ["server", "--no-token", "--host", "0.0.0.0", "--port", `${PORT}`],
-  Env: [
-    process.env.ANTHROPIC_API_KEY ? `ANTHROPIC_API_KEY=${process.env.ANTHROPIC_API_KEY}` : "",
-    process.env.OPENAI_API_KEY ? `OPENAI_API_KEY=${process.env.OPENAI_API_KEY}` : "",
-    process.env.CODEX_API_KEY ? `CODEX_API_KEY=${process.env.CODEX_API_KEY}` : "",
-  ].filter(Boolean),
-  ExposedPorts: { [`${PORT}/tcp`]: {} },
-  HostConfig: {
-    AutoRemove: true,
-    PortBindings: { [`${PORT}/tcp`]: [{ HostPort: `${PORT}` }] },
-    Binds: bindMounts,
-  },
 });
-await container.start();
 
-const baseUrl = `http://127.0.0.1:${PORT}`;
+console.log(`UI: ${client.inspectorUrl}`);
 
-const client = await SandboxAgent.connect({ baseUrl });
+const session = await client.createSession({
-const session = await client.createSession({ agent, sessionInit: { cwd: "/home/sandbox", mcpServers: [] } });
+  agent: detectAgent(),
-const sessionId = session.id;
+  cwd: "/home/sandbox",
+});
 
-console.log(`  UI: ${buildInspectorUrl({ baseUrl, sessionId })}`);
+session.onEvent((event) => {
-console.log("  Press Ctrl+C to stop.");
+  console.log(`[${event.sender}]`, JSON.stringify(event.payload));
+});
 
-const keepAlive = setInterval(() => {}, 60_000);
+session.prompt([{ type: "text", text: "Say hello from Docker in one sentence." }]);
-const cleanup = async () => {
+
-  clearInterval(keepAlive);
+process.once("SIGINT", async () => {
-  try {
+  await client.destroySandbox();
-    await container.stop({ t: 5 });
-  } catch {}
-  try {
-    await container.remove({ force: true });
-  } catch {}
   process.exit(0);
-};
+});
-process.once("SIGINT", cleanup);
-process.once("SIGTERM", cleanup);

examples/docker/tests/docker.test.ts

@@ -1,8 +1,15 @@
 import { describe, it, expect } from "vitest";
-import { buildHeaders } from "@sandbox-agent/example-shared";
+import { startDockerSandbox } from "@sandbox-agent/example-shared/docker";
-import { setupDockerSandboxAgent } from "../src/docker.ts";
 
-const shouldRun = process.env.RUN_DOCKER_EXAMPLES === "1";
+/**
+ * Docker integration test.
+ *
+ * Set SANDBOX_AGENT_DOCKER_IMAGE to the image tag to test (e.g. a locally-built
+ * full image). The test starts a container from that image, waits for
+ * sandbox-agent to become healthy, and validates the /v1/health endpoint.
+ */
+const image = process.env.SANDBOX_AGENT_DOCKER_IMAGE;
+const shouldRun = Boolean(image);
 const timeoutMs = Number.parseInt(process.env.SANDBOX_TEST_TIMEOUT_MS || "", 10) || 300_000;
 
 const testFn = shouldRun ? it : it.skip;
@@ -11,11 +18,29 @@ describe("docker example", () => {
   testFn(
     "starts sandbox-agent and responds to /v1/health",
     async () => {
-      const { baseUrl, token, cleanup } = await setupDockerSandboxAgent();
+      const { baseUrl, cleanup } = await startDockerSandbox({
+        port: 2468,
+        image: image!,
+      });
       try {
-        const response = await fetch(`${baseUrl}/v1/health`, {
+        // Wait for health check
-          headers: buildHeaders({ token }),
+        let healthy = false;
-        });
+        for (let i = 0; i < 60; i++) {
+          try {
+            const res = await fetch(`${baseUrl}/v1/health`);
+            if (res.ok) {
+              const data = await res.json();
+              if (data.status === "ok") {
+                healthy = true;
+                break;
+              }
+            }
+          } catch {}
+          await new Promise((r) => setTimeout(r, 1000));
+        }
+        expect(healthy).toBe(true);
+
+        const response = await fetch(`${baseUrl}/v1/health`);
         expect(response.ok).toBe(true);
         const data = await response.json();
         expect(data.status).toBe("ok");

examples/docker/tsconfig.json

@@ -9,7 +9,8 @@
     "esModuleInterop": true,
     "strict": true,
     "skipLibCheck": true,
-    "resolveJsonModule": true
+    "resolveJsonModule": true,
+    "types": ["node"]
   },
   "include": ["src/**/*"],
   "exclude": ["node_modules", "**/*.test.ts"]

examples/e2b/src/e2b.ts

@@ -0,0 +1,34 @@
+import { SandboxAgent } from "sandbox-agent";
+import { e2b } from "sandbox-agent/e2b";
+
+function collectEnvVars(): Record<string, string> {
+  const envs: Record<string, string> = {};
+  if (process.env.ANTHROPIC_API_KEY) envs.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
+  if (process.env.OPENAI_API_KEY) envs.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+  return envs;
+}
+
+function inspectorUrlToBaseUrl(inspectorUrl: string): string {
+  return inspectorUrl.replace(/\/ui\/$/, "");
+}
+
+export async function setupE2BSandboxAgent(): Promise<{
+  baseUrl: string;
+  token?: string;
+  cleanup: () => Promise<void>;
+}> {
+  const template = process.env.E2B_TEMPLATE;
+  const client = await SandboxAgent.start({
+    sandbox: e2b({
+      template,
+      create: { envs: collectEnvVars() },
+    }),
+  });
+
+  return {
+    baseUrl: inspectorUrlToBaseUrl(client.inspectorUrl),
+    cleanup: async () => {
+      await client.killSandbox();
+    },
+  };
+}

examples/e2b/src/index.ts

@@ -1,45 +1,28 @@
-import { Sandbox } from "@e2b/code-interpreter";
 import { SandboxAgent } from "sandbox-agent";
-import { detectAgent, buildInspectorUrl } from "@sandbox-agent/example-shared";
+import { e2b } from "sandbox-agent/e2b";
+import { detectAgent } from "@sandbox-agent/example-shared";
 
 const envs: Record<string, string> = {};
 if (process.env.ANTHROPIC_API_KEY) envs.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
 if (process.env.OPENAI_API_KEY) envs.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+const template = process.env.E2B_TEMPLATE;
 
-console.log("Creating E2B sandbox...");
+const client = await SandboxAgent.start({
-const sandbox = await Sandbox.create({ allowInternetAccess: true, envs });
+  // ✨ NEW ✨
+  sandbox: e2b({ template, create: { envs } }),
+});
 
-const run = async (cmd: string) => {
+const session = await client.createSession({
-  const result = await sandbox.commands.run(cmd);
+  agent: detectAgent(),
-  if (result.exitCode !== 0) throw new Error(`Command failed: ${cmd}\n${result.stderr}`);
+});
-  return result;
-};
 
-console.log("Installing sandbox-agent...");
+session.onEvent((event) => {
-await run("curl -fsSL https://releases.rivet.dev/sandbox-agent/0.3.x/install.sh | sh");
+  console.log(`[${event.sender}]`, JSON.stringify(event.payload));
+});
 
-console.log("Installing agents...");
+session.prompt([{ type: "text", text: "Say hello from E2B in one sentence." }]);
-await run("sandbox-agent install-agent claude");
-await run("sandbox-agent install-agent codex");
 
-console.log("Starting server...");
+process.once("SIGINT", async () => {
-await sandbox.commands.run("sandbox-agent server --no-token --host 0.0.0.0 --port 3000", { background: true, timeoutMs: 0 });
+  await client.destroySandbox();
-
-const baseUrl = `https://${sandbox.getHost(3000)}`;
-
-console.log("Connecting to server...");
-const client = await SandboxAgent.connect({ baseUrl });
-const session = await client.createSession({ agent: detectAgent(), sessionInit: { cwd: "/home/user", mcpServers: [] } });
-const sessionId = session.id;
-
-console.log(`  UI: ${buildInspectorUrl({ baseUrl, sessionId })}`);
-console.log("  Press Ctrl+C to stop.");
-
-const keepAlive = setInterval(() => {}, 60_000);
-const cleanup = async () => {
-  clearInterval(keepAlive);
-  await sandbox.kill();
   process.exit(0);
-};
+});
-process.once("SIGINT", cleanup);
-process.once("SIGTERM", cleanup);

examples/e2b/tsconfig.json

@@ -9,7 +9,8 @@
     "esModuleInterop": true,
     "strict": true,
     "skipLibCheck": true,
-    "resolveJsonModule": true
+    "resolveJsonModule": true,
+    "types": ["node"]
   },
   "include": ["src/**/*"],
   "exclude": ["node_modules", "**/*.test.ts"]

examples/file-system/src/index.ts

@@ -44,7 +44,7 @@ const readmeText = new TextDecoder().decode(readmeBytes);
 console.log(`  README.md content: ${readmeText.trim()}`);
 
 console.log("Creating session...");
-const session = await client.createSession({ agent: detectAgent(), sessionInit: { cwd: "/opt/my-project", mcpServers: [] } });
+const session = await client.createSession({ agent: detectAgent(), cwd: "/opt/my-project" });
 const sessionId = session.id;
 console.log(`  UI: ${buildInspectorUrl({ baseUrl, sessionId })}`);
 console.log('  Try: "read the README in /opt/my-project"');

examples/file-system/tsconfig.json

@@ -9,7 +9,8 @@
     "esModuleInterop": true,
     "strict": true,
     "skipLibCheck": true,
-    "resolveJsonModule": true
+    "resolveJsonModule": true,
+    "types": ["node"]
   },
   "include": ["src/**/*"],
   "exclude": ["node_modules", "**/*.test.ts"]

examples/mcp-custom-tool/tsconfig.json

@@ -9,7 +9,8 @@
     "esModuleInterop": true,
     "strict": true,
     "skipLibCheck": true,
-    "resolveJsonModule": true
+    "resolveJsonModule": true,
+    "types": ["node"]
   },
   "include": ["src/**/*"],
   "exclude": ["node_modules", "**/*.test.ts"]

examples/mcp/tsconfig.json

@@ -9,7 +9,8 @@
     "esModuleInterop": true,
     "strict": true,
     "skipLibCheck": true,
-    "resolveJsonModule": true
+    "resolveJsonModule": true,
+    "types": ["node"]
   },
   "include": ["src/**/*"],
   "exclude": ["node_modules", "**/*.test.ts"]

examples/modal/package.json

@@ -0,0 +1,20 @@
+{
+  "name": "@sandbox-agent/example-modal",
+  "private": true,
+  "type": "module",
+  "scripts": {
+    "start": "tsx src/index.ts",
+    "typecheck": "tsc --noEmit"
+  },
+  "dependencies": {
+    "modal": "latest",
+    "@sandbox-agent/example-shared": "workspace:*",
+    "sandbox-agent": "workspace:*"
+  },
+  "devDependencies": {
+    "@types/node": "latest",
+    "tsx": "latest",
+    "typescript": "latest",
+    "vitest": "^3.0.0"
+  }
+}

examples/modal/src/index.ts

@@ -0,0 +1,30 @@
+import { SandboxAgent } from "sandbox-agent";
+import { modal } from "sandbox-agent/modal";
+import { detectAgent } from "@sandbox-agent/example-shared";
+
+const secrets: Record<string, string> = {};
+if (process.env.ANTHROPIC_API_KEY) secrets.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
+if (process.env.OPENAI_API_KEY) secrets.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+
+const client = await SandboxAgent.start({
+  sandbox: modal({
+    create: { secrets },
+  }),
+});
+
+console.log(`UI: ${client.inspectorUrl}`);
+
+const session = await client.createSession({
+  agent: detectAgent(),
+});
+
+session.onEvent((event) => {
+  console.log(`[${event.sender}]`, JSON.stringify(event.payload));
+});
+
+session.prompt([{ type: "text", text: "Say hello from Modal in one sentence." }]);
+
+process.once("SIGINT", async () => {
+  await client.destroySandbox();
+  process.exit(0);
+});

examples/modal/tests/modal.test.ts

@@ -0,0 +1,31 @@
+import { describe, it, expect } from "vitest";
+import { SandboxAgent } from "sandbox-agent";
+import { modal } from "sandbox-agent/modal";
+
+const shouldRun = Boolean(process.env.MODAL_TOKEN_ID && process.env.MODAL_TOKEN_SECRET);
+const timeoutMs = Number.parseInt(process.env.SANDBOX_TEST_TIMEOUT_MS || "", 10) || 300_000;
+
+const testFn = shouldRun ? it : it.skip;
+
+describe("modal provider", () => {
+  testFn(
+    "starts sandbox-agent and responds to /v1/health",
+    async () => {
+      const secrets: Record<string, string> = {};
+      if (process.env.ANTHROPIC_API_KEY) secrets.ANTHROPIC_API_KEY = process.env.ANTHROPIC_API_KEY;
+      if (process.env.OPENAI_API_KEY) secrets.OPENAI_API_KEY = process.env.OPENAI_API_KEY;
+
+      const sdk = await SandboxAgent.start({
+        sandbox: modal({ create: { secrets } }),
+      });
+
+      try {
+        const health = await sdk.getHealth();
+        expect(health.status).toBe("ok");
+      } finally {
+        await sdk.destroySandbox();
+      }
+    },
+    timeoutMs,
+  );
+});

examples/modal/tsconfig.json

@@ -0,0 +1,17 @@
+{
+  "compilerOptions": {
+    "target": "ES2022",
+    "lib": ["ES2022", "DOM"],
+    "module": "ESNext",
+    "moduleResolution": "Bundler",
+    "allowImportingTsExtensions": true,
+    "noEmit": true,
+    "esModuleInterop": true,
+    "strict": true,
+    "skipLibCheck": true,
+    "resolveJsonModule": true,
+    "types": ["node"]
+  },
+  "include": ["src/**/*"],
+  "exclude": ["node_modules", "**/*.test.ts"]
+}

examples/permissions/src/index.ts

@@ -2,6 +2,7 @@ import { createInterface } from "node:readline/promises";
 import { stdin as input, stdout as output } from "node:process";
 import { Command } from "commander";
 import { SandboxAgent, type PermissionReply, type SessionPermissionRequest } from "sandbox-agent";
+import { local } from "sandbox-agent/local";
 
 const options = parseOptions();
 const agent = options.agent.trim().toLowerCase();
@@ -9,10 +10,7 @@ const autoReply = parsePermissionReply(options.reply);
 const promptText = options.prompt?.trim() || `Create ./permission-example.txt with the text 'hello from the ${agent} permissions example'.`;
 
 const sdk = await SandboxAgent.start({
-  spawn: {
+  sandbox: local({ log: "inherit" }),
-    enabled: true,
-    log: "inherit",
-  },
 });
 
 try {
@@ -43,10 +41,7 @@ try {
   const session = await sdk.createSession({
     agent,
     ...(mode ? { mode } : {}),
-    sessionInit: {
+    cwd: process.cwd(),
-      cwd: process.cwd(),
-      mcpServers: [],
-    },
   });
 
   const rl = autoReply

examples/permissions/tsconfig.json

@@ -1,7 +1,8 @@
 {
   "compilerOptions": {
     "target": "ES2022",
-    "lib": ["ES2022"],
+    "lib": ["ES2022", "DOM"],
+    "types": ["node"],
     "module": "ESNext",
     "moduleResolution": "Bundler",
     "allowImportingTsExtensions": true,

examples/persist-memory/tsconfig.json

@@ -1,13 +1,15 @@
 {
   "compilerOptions": {
     "target": "ES2022",
+    "lib": ["ES2022", "DOM"],
     "module": "ESNext",
     "moduleResolution": "Bundler",
     "allowImportingTsExtensions": true,
     "noEmit": true,
     "esModuleInterop": true,
     "strict": true,
-    "skipLibCheck": true
+    "skipLibCheck": true,
+    "types": ["node"]
   },
   "include": ["src"]
 }

examples/persist-postgres/package.json

@@ -8,7 +8,6 @@
   },
   "dependencies": {
     "@sandbox-agent/example-shared": "workspace:*",
-    "@sandbox-agent/persist-postgres": "workspace:*",
     "pg": "latest",
     "sandbox-agent": "workspace:*"
   },

examples/persist-postgres/src/index.ts

@@ -3,7 +3,7 @@ import { randomUUID } from "node:crypto";
 import { Client } from "pg";
 import { setTimeout as delay } from "node:timers/promises";
 import { SandboxAgent } from "sandbox-agent";
-import { PostgresSessionPersistDriver } from "@sandbox-agent/persist-postgres";
+import { PostgresSessionPersistDriver } from "./persist.ts";
 import { startDockerSandbox } from "@sandbox-agent/example-shared/docker";
 import { detectAgent } from "@sandbox-agent/example-shared";
 

examples/persist-postgres/src/persist.ts

@@ -0,0 +1,336 @@
+import { Pool, type PoolConfig } from "pg";
+import type { ListEventsRequest, ListPage, ListPageRequest, SessionEvent, SessionPersistDriver, SessionRecord } from "sandbox-agent";
+
+const DEFAULT_LIST_LIMIT = 100;
+
+export interface PostgresSessionPersistDriverOptions {
+  connectionString?: string;
+  pool?: Pool;
+  poolConfig?: PoolConfig;
+  schema?: string;
+}
+
+export class PostgresSessionPersistDriver implements SessionPersistDriver {
+  private readonly pool: Pool;
+  private readonly ownsPool: boolean;
+  private readonly schema: string;
+  private readonly initialized: Promise<void>;
+
+  constructor(options: PostgresSessionPersistDriverOptions = {}) {
+    this.schema = normalizeSchema(options.schema ?? "public");
+
+    if (options.pool) {
+      this.pool = options.pool;
+      this.ownsPool = false;
+    } else {
+      this.pool = new Pool({
+        connectionString: options.connectionString,
+        ...options.poolConfig,
+      });
+      this.ownsPool = true;
+    }
+
+    this.initialized = this.initialize();
+  }
+
+  async getSession(id: string): Promise<SessionRecord | undefined> {
+    await this.ready();
+
+    const result = await this.pool.query<SessionRow>(
+      `SELECT id, agent, agent_session_id, last_connection_id, created_at, destroyed_at, sandbox_id, session_init_json, config_options_json, modes_json
+       FROM ${this.table("sessions")}
+       WHERE id = $1`,
+      [id],
+    );
+
+    if (result.rows.length === 0) {
+      return undefined;
+    }
+
+    return decodeSessionRow(result.rows[0]);
+  }
+
+  async listSessions(request: ListPageRequest = {}): Promise<ListPage<SessionRecord>> {
+    await this.ready();
+
+    const offset = parseCursor(request.cursor);
+    const limit = normalizeLimit(request.limit);
+
+    const rowsResult = await this.pool.query<SessionRow>(
+      `SELECT id, agent, agent_session_id, last_connection_id, created_at, destroyed_at, sandbox_id, session_init_json, config_options_json, modes_json
+       FROM ${this.table("sessions")}
+       ORDER BY created_at ASC, id ASC
+       LIMIT $1 OFFSET $2`,
+      [limit, offset],
+    );
+
+    const countResult = await this.pool.query<{ count: string }>(`SELECT COUNT(*) AS count FROM ${this.table("sessions")}`);
+    const total = parseInteger(countResult.rows[0]?.count ?? "0");
+    const nextOffset = offset + rowsResult.rows.length;
+
+    return {
+      items: rowsResult.rows.map(decodeSessionRow),
+      nextCursor: nextOffset < total ? String(nextOffset) : undefined,
+    };
+  }
+
+  async updateSession(session: SessionRecord): Promise<void> {
+    await this.ready();
+
+    await this.pool.query(
+      `INSERT INTO ${this.table("sessions")} (
+        id, agent, agent_session_id, last_connection_id, created_at, destroyed_at, sandbox_id, session_init_json, config_options_json, modes_json
+      ) VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10)
+      ON CONFLICT(id) DO UPDATE SET
+        agent = EXCLUDED.agent,
+        agent_session_id = EXCLUDED.agent_session_id,
+        last_connection_id = EXCLUDED.last_connection_id,
+        created_at = EXCLUDED.created_at,
+        destroyed_at = EXCLUDED.destroyed_at,
+        sandbox_id = EXCLUDED.sandbox_id,
+        session_init_json = EXCLUDED.session_init_json,
+        config_options_json = EXCLUDED.config_options_json,
+        modes_json = EXCLUDED.modes_json`,
+      [
+        session.id,
+        session.agent,
+        session.agentSessionId,
+        session.lastConnectionId,
+        session.createdAt,
+        session.destroyedAt ?? null,
+        session.sandboxId ?? null,
+        session.sessionInit ? JSON.stringify(session.sessionInit) : null,
+        session.configOptions ? JSON.stringify(session.configOptions) : null,
+        session.modes !== undefined ? JSON.stringify(session.modes) : null,
+      ],
+    );
+  }
+
+  async listEvents(request: ListEventsRequest): Promise<ListPage<SessionEvent>> {
+    await this.ready();
+
+    const offset = parseCursor(request.cursor);
+    const limit = normalizeLimit(request.limit);
+
+    const rowsResult = await this.pool.query<EventRow>(
+      `SELECT id, event_index, session_id, created_at, connection_id, sender, payload_json
+       FROM ${this.table("events")}
+       WHERE session_id = $1
+       ORDER BY event_index ASC, id ASC
+       LIMIT $2 OFFSET $3`,
+      [request.sessionId, limit, offset],
+    );
+
+    const countResult = await this.pool.query<{ count: string }>(`SELECT COUNT(*) AS count FROM ${this.table("events")} WHERE session_id = $1`, [
+      request.sessionId,
+    ]);
+    const total = parseInteger(countResult.rows[0]?.count ?? "0");
+    const nextOffset = offset + rowsResult.rows.length;
+
+    return {
+      items: rowsResult.rows.map(decodeEventRow),
+      nextCursor: nextOffset < total ? String(nextOffset) : undefined,
+    };
+  }
+
+  async insertEvent(_sessionId: string, event: SessionEvent): Promise<void> {
+    await this.ready();
+
+    await this.pool.query(
+      `INSERT INTO ${this.table("events")} (
+        id, event_index, session_id, created_at, connection_id, sender, payload_json
+      ) VALUES ($1, $2, $3, $4, $5, $6, $7)
+      ON CONFLICT(id) DO UPDATE SET
+        event_index = EXCLUDED.event_index,
+        session_id = EXCLUDED.session_id,
+        created_at = EXCLUDED.created_at,
+        connection_id = EXCLUDED.connection_id,
+        sender = EXCLUDED.sender,
+        payload_json = EXCLUDED.payload_json`,
+      [event.id, event.eventIndex, event.sessionId, event.createdAt, event.connectionId, event.sender, event.payload],
+    );
+  }
+
+  async close(): Promise<void> {
+    if (!this.ownsPool) {
+      return;
+    }
+    await this.pool.end();
+  }
+
+  private async ready(): Promise<void> {
+    await this.initialized;
+  }
+
+  private table(name: "sessions" | "events"): string {
+    return `"${this.schema}"."${name}"`;
+  }
+
+  private async initialize(): Promise<void> {
+    await this.pool.query(`CREATE SCHEMA IF NOT EXISTS "${this.schema}"`);
+
+    await this.pool.query(`
+      CREATE TABLE IF NOT EXISTS ${this.table("sessions")} (
+        id TEXT PRIMARY KEY,
+        agent TEXT NOT NULL,
+        agent_session_id TEXT NOT NULL,
+        last_connection_id TEXT NOT NULL,
+        created_at BIGINT NOT NULL,
+        destroyed_at BIGINT,
+        sandbox_id TEXT,
+        session_init_json JSONB,
+        config_options_json JSONB,
+        modes_json JSONB
+      )
+    `);
+
+    await this.pool.query(`
+      ALTER TABLE ${this.table("sessions")}
+      ADD COLUMN IF NOT EXISTS sandbox_id TEXT
+    `);
+
+    await this.pool.query(`
+      ALTER TABLE ${this.table("sessions")}
+      ADD COLUMN IF NOT EXISTS config_options_json JSONB
+    `);
+
+    await this.pool.query(`
+      ALTER TABLE ${this.table("sessions")}
+      ADD COLUMN IF NOT EXISTS modes_json JSONB
+    `);
+
+    await this.pool.query(`
+      CREATE TABLE IF NOT EXISTS ${this.table("events")} (
+        id TEXT PRIMARY KEY,
+        event_index BIGINT NOT NULL,
+        session_id TEXT NOT NULL,
+        created_at BIGINT NOT NULL,
+        connection_id TEXT NOT NULL,
+        sender TEXT NOT NULL,
+        payload_json JSONB NOT NULL
+      )
+    `);
+
+    await this.pool.query(`
+      ALTER TABLE ${this.table("events")}
+      ALTER COLUMN id TYPE TEXT USING id::TEXT
+    `);
+
+    await this.pool.query(`
+      ALTER TABLE ${this.table("events")}
+      ADD COLUMN IF NOT EXISTS event_index BIGINT
+    `);
+
+    await this.pool.query(`
+      WITH ranked AS (
+        SELECT id, ROW_NUMBER() OVER (PARTITION BY session_id ORDER BY created_at ASC, id ASC) AS ranked_index
+        FROM ${this.table("events")}
+      )
+      UPDATE ${this.table("events")} AS current_events
+      SET event_index = ranked.ranked_index
+      FROM ranked
+      WHERE current_events.id = ranked.id
+        AND current_events.event_index IS NULL
+    `);
+
+    await this.pool.query(`
+      ALTER TABLE ${this.table("events")}
+      ALTER COLUMN event_index SET NOT NULL
+    `);
+
+    await this.pool.query(`
+      CREATE INDEX IF NOT EXISTS idx_events_session_order
+      ON ${this.table("events")}(session_id, event_index, id)
+    `);
+  }
+}
+
+type SessionRow = {
+  id: string;
+  agent: string;
+  agent_session_id: string;
+  last_connection_id: string;
+  created_at: string | number;
+  destroyed_at: string | number | null;
+  sandbox_id: string | null;
+  session_init_json: unknown | null;
+  config_options_json: unknown | null;
+  modes_json: unknown | null;
+};
+
+type EventRow = {
+  id: string | number;
+  event_index: string | number;
+  session_id: string;
+  created_at: string | number;
+  connection_id: string;
+  sender: string;
+  payload_json: unknown;
+};
+
+function decodeSessionRow(row: SessionRow): SessionRecord {
+  return {
+    id: row.id,
+    agent: row.agent,
+    agentSessionId: row.agent_session_id,
+    lastConnectionId: row.last_connection_id,
+    createdAt: parseInteger(row.created_at),
+    destroyedAt: row.destroyed_at === null ? undefined : parseInteger(row.destroyed_at),
+    sandboxId: row.sandbox_id ?? undefined,
+    sessionInit: row.session_init_json ? (row.session_init_json as SessionRecord["sessionInit"]) : undefined,
+    configOptions: row.config_options_json ? (row.config_options_json as SessionRecord["configOptions"]) : undefined,
+    modes: row.modes_json ? (row.modes_json as SessionRecord["modes"]) : undefined,
+  };
+}
+
+function decodeEventRow(row: EventRow): SessionEvent {
+  return {
+    id: String(row.id),
+    eventIndex: parseInteger(row.event_index),
+    sessionId: row.session_id,
+    createdAt: parseInteger(row.created_at),
+    connectionId: row.connection_id,
+    sender: parseSender(row.sender),
+    payload: row.payload_json as SessionEvent["payload"],
+  };
+}
+
+function normalizeLimit(limit: number | undefined): number {
+  if (!Number.isFinite(limit) || (limit ?? 0) < 1) {
+    return DEFAULT_LIST_LIMIT;
+  }
+  return Math.floor(limit as number);
+}
+
+function parseCursor(cursor: string | undefined): number {
+  if (!cursor) {
+    return 0;
+  }
+  const parsed = Number.parseInt(cursor, 10);
+  if (!Number.isFinite(parsed) || parsed < 0) {
+    return 0;
+  }
+  return parsed;
+}
+
+function parseInteger(value: string | number): number {
+  const parsed = typeof value === "number" ? value : Number.parseInt(value, 10);
+  if (!Number.isFinite(parsed)) {
+    throw new Error(`Invalid integer value returned by postgres: ${String(value)}`);
+  }
+  return parsed;
+}
+
+function parseSender(value: string): SessionEvent["sender"] {
+  if (value === "agent" || value === "client") {
+    return value;
+  }
+  throw new Error(`Invalid sender value returned by postgres: ${value}`);
+}
+
+function normalizeSchema(schema: string): string {
+  if (!/^[A-Za-z_][A-Za-z0-9_]*$/.test(schema)) {
+    throw new Error(`Invalid schema name '${schema}'. Use letters, numbers, and underscores only.`);
+  }
+  return schema;
+}

examples/persist-postgres/tsconfig.json

@@ -1,13 +1,15 @@
 {
   "compilerOptions": {
     "target": "ES2022",
+    "lib": ["ES2022", "DOM"],
     "module": "ESNext",
     "moduleResolution": "Bundler",
     "allowImportingTsExtensions": true,
     "noEmit": true,
     "esModuleInterop": true,
     "strict": true,
-    "skipLibCheck": true
+    "skipLibCheck": true,
+    "types": ["node"]
   },
   "include": ["src"]
 }

examples/persist-sqlite/package.json

@@ -8,10 +8,11 @@
   },
   "dependencies": {
     "@sandbox-agent/example-shared": "workspace:*",
-    "@sandbox-agent/persist-sqlite": "workspace:*",
+    "better-sqlite3": "^11.0.0",
     "sandbox-agent": "workspace:*"
   },
   "devDependencies": {
+    "@types/better-sqlite3": "^7.0.0",
     "@types/node": "latest",
     "tsx": "latest",
     "typescript": "latest"

examples/persist-sqlite/src/index.ts

@@ -1,5 +1,5 @@
 import { SandboxAgent } from "sandbox-agent";
-import { SQLiteSessionPersistDriver } from "@sandbox-agent/persist-sqlite";
+import { SQLiteSessionPersistDriver } from "./persist.ts";
 import { startDockerSandbox } from "@sandbox-agent/example-shared/docker";
 import { detectAgent } from "@sandbox-agent/example-shared";
 

examples/persist-sqlite/src/persist.ts

@@ -0,0 +1,310 @@
+import Database from "better-sqlite3";
+import type { ListEventsRequest, ListPage, ListPageRequest, SessionEvent, SessionPersistDriver, SessionRecord } from "sandbox-agent";
+
+const DEFAULT_LIST_LIMIT = 100;
+
+export interface SQLiteSessionPersistDriverOptions {
+  filename?: string;
+}
+
+export class SQLiteSessionPersistDriver implements SessionPersistDriver {
+  private readonly db: Database.Database;
+
+  constructor(options: SQLiteSessionPersistDriverOptions = {}) {
+    this.db = new Database(options.filename ?? ":memory:");
+    this.initialize();
+  }
+
+  async getSession(id: string): Promise<SessionRecord | undefined> {
+    const row = this.db
+      .prepare(
+        `SELECT id, agent, agent_session_id, last_connection_id, created_at, destroyed_at, sandbox_id, session_init_json, config_options_json, modes_json
+         FROM sessions WHERE id = ?`,
+      )
+      .get(id) as SessionRow | undefined;
+
+    if (!row) {
+      return undefined;
+    }
+
+    return decodeSessionRow(row);
+  }
+
+  async listSessions(request: ListPageRequest = {}): Promise<ListPage<SessionRecord>> {
+    const offset = parseCursor(request.cursor);
+    const limit = normalizeLimit(request.limit);
+
+    const rows = this.db
+      .prepare(
+        `SELECT id, agent, agent_session_id, last_connection_id, created_at, destroyed_at, sandbox_id, session_init_json, config_options_json, modes_json
+         FROM sessions
+         ORDER BY created_at ASC, id ASC
+         LIMIT ? OFFSET ?`,
+      )
+      .all(limit, offset) as SessionRow[];
+
+    const countRow = this.db.prepare(`SELECT COUNT(*) as count FROM sessions`).get() as { count: number };
+    const nextOffset = offset + rows.length;
+
+    return {
+      items: rows.map(decodeSessionRow),
+      nextCursor: nextOffset < countRow.count ? String(nextOffset) : undefined,
+    };
+  }
+
+  async updateSession(session: SessionRecord): Promise<void> {
+    this.db
+      .prepare(
+        `INSERT INTO sessions (
+          id, agent, agent_session_id, last_connection_id, created_at, destroyed_at, sandbox_id, session_init_json, config_options_json, modes_json
+        ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+        ON CONFLICT(id) DO UPDATE SET
+          agent = excluded.agent,
+          agent_session_id = excluded.agent_session_id,
+          last_connection_id = excluded.last_connection_id,
+          created_at = excluded.created_at,
+          destroyed_at = excluded.destroyed_at,
+          sandbox_id = excluded.sandbox_id,
+          session_init_json = excluded.session_init_json,
+          config_options_json = excluded.config_options_json,
+          modes_json = excluded.modes_json`,
+      )
+      .run(
+        session.id,
+        session.agent,
+        session.agentSessionId,
+        session.lastConnectionId,
+        session.createdAt,
+        session.destroyedAt ?? null,
+        session.sandboxId ?? null,
+        session.sessionInit ? JSON.stringify(session.sessionInit) : null,
+        session.configOptions ? JSON.stringify(session.configOptions) : null,
+        session.modes !== undefined ? JSON.stringify(session.modes) : null,
+      );
+  }
+
+  async listEvents(request: ListEventsRequest): Promise<ListPage<SessionEvent>> {
+    const offset = parseCursor(request.cursor);
+    const limit = normalizeLimit(request.limit);
+
+    const rows = this.db
+      .prepare(
+        `SELECT id, event_index, session_id, created_at, connection_id, sender, payload_json
+         FROM events
+         WHERE session_id = ?
+         ORDER BY event_index ASC, id ASC
+         LIMIT ? OFFSET ?`,
+      )
+      .all(request.sessionId, limit, offset) as EventRow[];
+
+    const countRow = this.db.prepare(`SELECT COUNT(*) as count FROM events WHERE session_id = ?`).get(request.sessionId) as { count: number };
+
+    const nextOffset = offset + rows.length;
+
+    return {
+      items: rows.map(decodeEventRow),
+      nextCursor: nextOffset < countRow.count ? String(nextOffset) : undefined,
+    };
+  }
+
+  async insertEvent(_sessionId: string, event: SessionEvent): Promise<void> {
+    this.db
+      .prepare(
+        `INSERT INTO events (
+          id, event_index, session_id, created_at, connection_id, sender, payload_json
+        ) VALUES (?, ?, ?, ?, ?, ?, ?)
+        ON CONFLICT(id) DO UPDATE SET
+          event_index = excluded.event_index,
+          session_id = excluded.session_id,
+          created_at = excluded.created_at,
+          connection_id = excluded.connection_id,
+          sender = excluded.sender,
+          payload_json = excluded.payload_json`,
+      )
+      .run(event.id, event.eventIndex, event.sessionId, event.createdAt, event.connectionId, event.sender, JSON.stringify(event.payload));
+  }
+
+  close(): void {
+    this.db.close();
+  }
+
+  private initialize(): void {
+    this.db.exec(`
+      CREATE TABLE IF NOT EXISTS sessions (
+        id TEXT PRIMARY KEY,
+        agent TEXT NOT NULL,
+        agent_session_id TEXT NOT NULL,
+        last_connection_id TEXT NOT NULL,
+        created_at INTEGER NOT NULL,
+        destroyed_at INTEGER,
+        sandbox_id TEXT,
+        session_init_json TEXT,
+        config_options_json TEXT,
+        modes_json TEXT
+      )
+    `);
+
+    const sessionColumns = this.db.prepare(`PRAGMA table_info(sessions)`).all() as TableInfoRow[];
+    if (!sessionColumns.some((column) => column.name === "sandbox_id")) {
+      this.db.exec(`ALTER TABLE sessions ADD COLUMN sandbox_id TEXT`);
+    }
+    if (!sessionColumns.some((column) => column.name === "config_options_json")) {
+      this.db.exec(`ALTER TABLE sessions ADD COLUMN config_options_json TEXT`);
+    }
+    if (!sessionColumns.some((column) => column.name === "modes_json")) {
+      this.db.exec(`ALTER TABLE sessions ADD COLUMN modes_json TEXT`);
+    }
+
+    this.ensureEventsTable();
+  }
+
+  private ensureEventsTable(): void {
+    const tableInfo = this.db.prepare(`PRAGMA table_info(events)`).all() as TableInfoRow[];
+    if (tableInfo.length === 0) {
+      this.createEventsTable();
+      return;
+    }
+
+    const idColumn = tableInfo.find((column) => column.name === "id");
+    const hasEventIndex = tableInfo.some((column) => column.name === "event_index");
+    const idType = (idColumn?.type ?? "").trim().toUpperCase();
+    const idIsText = idType === "TEXT";
+
+    if (!idIsText || !hasEventIndex) {
+      this.rebuildEventsTable(hasEventIndex);
+    }
+
+    this.db.exec(`
+      CREATE INDEX IF NOT EXISTS idx_events_session_order
+      ON events(session_id, event_index, id)
+    `);
+  }
+
+  private createEventsTable(): void {
+    this.db.exec(`
+      CREATE TABLE IF NOT EXISTS events (
+        id TEXT PRIMARY KEY,
+        event_index INTEGER NOT NULL,
+        session_id TEXT NOT NULL,
+        created_at INTEGER NOT NULL,
+        connection_id TEXT NOT NULL,
+        sender TEXT NOT NULL,
+        payload_json TEXT NOT NULL
+      );
+
+      CREATE INDEX IF NOT EXISTS idx_events_session_order
+      ON events(session_id, event_index, id)
+    `);
+  }
+
+  private rebuildEventsTable(hasEventIndex: boolean): void {
+    this.db.exec(`
+      ALTER TABLE events RENAME TO events_legacy;
+    `);
+
+    this.createEventsTable();
+
+    if (hasEventIndex) {
+      this.db.exec(`
+        INSERT INTO events (id, event_index, session_id, created_at, connection_id, sender, payload_json)
+        SELECT
+          CAST(id AS TEXT),
+          COALESCE(event_index, ROW_NUMBER() OVER (PARTITION BY session_id ORDER BY created_at ASC, id ASC)),
+          session_id,
+          created_at,
+          connection_id,
+          sender,
+          payload_json
+        FROM events_legacy
+      `);
+    } else {
+      this.db.exec(`
+        INSERT INTO events (id, event_index, session_id, created_at, connection_id, sender, payload_json)
+        SELECT
+          CAST(id AS TEXT),
+          ROW_NUMBER() OVER (PARTITION BY session_id ORDER BY created_at ASC, id ASC),
+          session_id,
+          created_at,
+          connection_id,
+          sender,
+          payload_json
+        FROM events_legacy
+      `);
+    }
+
+    this.db.exec(`DROP TABLE events_legacy`);
+  }
+}
+
+type SessionRow = {
+  id: string;
+  agent: string;
+  agent_session_id: string;
+  last_connection_id: string;
+  created_at: number;
+  destroyed_at: number | null;
+  sandbox_id: string | null;
+  session_init_json: string | null;
+  config_options_json: string | null;
+  modes_json: string | null;
+};
+
+type EventRow = {
+  id: string;
+  event_index: number;
+  session_id: string;
+  created_at: number;
+  connection_id: string;
+  sender: "client" | "agent";
+  payload_json: string;
+};
+
+type TableInfoRow = {
+  name: string;
+  type: string;
+};
+
+function decodeSessionRow(row: SessionRow): SessionRecord {
+  return {
+    id: row.id,
+    agent: row.agent,
+    agentSessionId: row.agent_session_id,
+    lastConnectionId: row.last_connection_id,
+    createdAt: row.created_at,
+    destroyedAt: row.destroyed_at ?? undefined,
+    sandboxId: row.sandbox_id ?? undefined,
+    sessionInit: row.session_init_json ? (JSON.parse(row.session_init_json) as SessionRecord["sessionInit"]) : undefined,
+    configOptions: row.config_options_json ? (JSON.parse(row.config_options_json) as SessionRecord["configOptions"]) : undefined,
+    modes: row.modes_json ? (JSON.parse(row.modes_json) as SessionRecord["modes"]) : undefined,
+  };
+}
+
+function decodeEventRow(row: EventRow): SessionEvent {
+  return {
+    id: row.id,
+    eventIndex: row.event_index,
+    sessionId: row.session_id,
+    createdAt: row.created_at,
+    connectionId: row.connection_id,
+    sender: row.sender,
+    payload: JSON.parse(row.payload_json),
+  };
+}
+
+function normalizeLimit(limit: number | undefined): number {
+  if (!Number.isFinite(limit) || (limit ?? 0) < 1) {
+    return DEFAULT_LIST_LIMIT;
+  }
+  return Math.floor(limit as number);
+}
+
+function parseCursor(cursor: string | undefined): number {
+  if (!cursor) {
+    return 0;
+  }
+  const parsed = Number.parseInt(cursor, 10);
+  if (!Number.isFinite(parsed) || parsed < 0) {
+    return 0;
+  }
+  return parsed;
+}

examples/persist-sqlite/tsconfig.json

@@ -1,13 +1,15 @@
 {
   "compilerOptions": {
     "target": "ES2022",
+    "lib": ["ES2022", "DOM"],
     "module": "ESNext",
     "moduleResolution": "Bundler",
     "allowImportingTsExtensions": true,
     "noEmit": true,
     "esModuleInterop": true,
     "strict": true,
-    "skipLibCheck": true
+    "skipLibCheck": true,
+    "types": ["node"]
   },
   "include": ["src"]
 }

examples/shared/src/docker.ts

@@ -9,7 +9,7 @@ const __dirname = path.dirname(fileURLToPath(import.meta.url));
 const REPO_ROOT = path.resolve(__dirname, "..", "..", "..");
 
 /** Pre-built Docker image with all agents installed. */
-export const FULL_IMAGE = "rivetdev/sandbox-agent:0.3.1-full";
+export const FULL_IMAGE = "rivetdev/sandbox-agent:0.4.2-full";
 
 export interface DockerSandboxOptions {
   /** Container port used by sandbox-agent inside Docker. */
@@ -78,11 +78,11 @@ function readClaudeCredentialFiles(): ClaudeCredentialFile[] {
   const candidates: Array<{ hostPath: string; containerPath: string }> = [
     {
       hostPath: path.join(homeDir, ".claude", ".credentials.json"),
-      containerPath: "/root/.claude/.credentials.json",
+      containerPath: ".claude/.credentials.json",
     },
     {
       hostPath: path.join(homeDir, ".claude-oauth-credentials.json"),
-      containerPath: "/root/.claude-oauth-credentials.json",
+      containerPath: ".claude-oauth-credentials.json",
     },
   ];
 
@@ -180,10 +180,9 @@ export async function startDockerSandbox(opts: DockerSandboxOptions): Promise<Do
     const credentialBootstrapCommands = claudeCredentialFiles.flatMap((file, index) => {
       const envKey = `SANDBOX_AGENT_CLAUDE_CREDENTIAL_${index}_B64`;
       bootstrapEnv[envKey] = file.base64Content;
-      return [
+      // Use $HOME-relative paths so credentials work regardless of container user
-        `mkdir -p ${shellSingleQuotedLiteral(path.posix.dirname(file.containerPath))}`,
+      const containerDir = path.posix.dirname(file.containerPath);
-        `printf %s "$${envKey}" | base64 -d > ${shellSingleQuotedLiteral(file.containerPath)}`,
+      return [`mkdir -p "$HOME/${containerDir}"`, `printf %s "$${envKey}" | base64 -d > "$HOME/${file.containerPath}"`];
-      ];
     });
     setupCommands.unshift(...credentialBootstrapCommands);
   }
@@ -200,8 +199,9 @@ export async function startDockerSandbox(opts: DockerSandboxOptions): Promise<Do
 
   const container = await docker.createContainer({
     Image: image,
+    Entrypoint: ["/bin/sh", "-c"],
     WorkingDir: "/home/sandbox",
-    Cmd: ["sh", "-c", bootCommands.join(" && ")],
+    Cmd: [bootCommands.join(" && ")],
     Env: [...Object.entries(credentialEnv).map(([key, value]) => `${key}=${value}`), ...Object.entries(bootstrapEnv).map(([key, value]) => `${key}=${value}`)],
     ExposedPorts: { [`${port}/tcp`]: {} },
     HostConfig: {
@@ -253,10 +253,13 @@ export async function startDockerSandbox(opts: DockerSandboxOptions): Promise<Do
     try {
       await container.remove({ force: true });
     } catch {}
+  };
+  const signalCleanup = async () => {
+    await cleanup();
     process.exit(0);
   };
-  process.once("SIGINT", cleanup);
+  process.once("SIGINT", signalCleanup);
-  process.once("SIGTERM", cleanup);
+  process.once("SIGTERM", signalCleanup);
 
   return { baseUrl, cleanup };
 }