harivansh-afk/sandbox-agent
1
0
Fork 0
mirror of https://github.com/harivansh-afk/sandbox-agent.git synced 2026-04-15 06:04:43 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Improve Daytona sandbox provisioning and frontend UI

Browse source 
Refactor git clone script in Daytona provider to use cleaner shell logic for GitHub token authentication and branch checkout. Add support for private repository clones with token-based auth. Improve Daytona provider error handling and git configuration setup.

Frontend improvements include enhanced dev panel, workspace dashboard, sidebar navigation, and UI components for better task/session management. Update interest manager and backend client to support improved session state handling.

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
This commit is contained in:
Nathan Flurry 2026-03-13 23:06:24 -07:00
parent 8fb19b50da
commit 098b8113f3
19 changed files with 394 additions and 130 deletions
Download patch file Download diff file Expand all files Collapse all files

2
foundry/CLAUDE.md
View file

@ -208,6 +208,8 @@ For all Rivet/RivetKit implementation:
- Read paths must not force refresh/sync work inline. Serve the latest cached projection, mark staleness explicitly, and trigger background refresh separately when needed.
- If a workflow needs to resume after some external work completes, model that as workflow state plus follow-up messages/events instead of holding the original request open.
- No retries: never add retry loops (`withRetries`, `setTimeout` retry, exponential backoff) anywhere in the codebase. If an operation fails, surface the error immediately. If a dependency is not ready yet, model that explicitly with workflow state and resume from a push/event instead of polling or retry loops.
- Never throw errors that expect the caller to retry (e.g. `throw new Error("... retry shortly")`). If a dependency is not ready, write the current state to the DB with an appropriate pending status, enqueue the async work, and return successfully. Let the client observe the pending → ready transition via push events.
- Action return contract: every action that creates a resource must write the resource record to the DB before returning, so the client can immediately query/render it. The record may have a pending status, but it must exist. Never return an ID that doesn't yet have a corresponding DB row.
- Actor handle policy:
- Prefer explicit `get` or explicit `create` based on workflow intent; do not default to `getOrCreate`.
- Use `get`/`getForId` when the actor is expected to already exist; if missing, surface an explicit `Actor not found` error with recovery context.

6
foundry/packages/backend/src/actors/project/actions.ts
View file

@ -437,7 +437,6 @@ async function hydrateTaskIndexMutation(c: any, _cmd?: HydrateTaskIndexCommand):
}
async function createTaskMutation(c: any, cmd: CreateTaskCommand): Promise<TaskRecord> {
const localPath = await ensureProjectReady(c);
const onBranch = cmd.onBranch?.trim() || null;
const initialBranchName = onBranch;
const initialTitle = onBranch ? deriveFallbackTitle(cmd.task, cmd.explicitTitle ?? undefined) : null;
@ -463,7 +462,6 @@ async function createTaskMutation(c: any, cmd: CreateTaskCommand): Promise<TaskR
repoId: c.state.repoId,
taskId,
repoRemote: c.state.remoteUrl,
repoLocalPath: localPath,
branchName: initialBranchName,
title: initialTitle,
task: cmd.task,
@ -954,7 +952,7 @@ export async function runProjectWorkflow(ctx: any): Promise<void> {
if (msg.name === "project.command.createTask") {
const result = await loopCtx.step({
name: "project-create-task",
timeout: 12 * 60_000,
timeout: 60_000,
run: async () => createTaskMutation(loopCtx, msg.body as CreateTaskCommand),
});
await msg.complete(result);
@ -1020,7 +1018,7 @@ export const projectActions = {
return expectQueueResponse<TaskRecord>(
await self.send(projectWorkflowQueueName("project.command.createTask"), cmd, {
wait: true,
timeout: 12 * 60_000,
timeout: 60_000,
}),
);
},

2
foundry/packages/backend/src/actors/task/index.ts
View file

@ -41,7 +41,7 @@ export interface TaskInput {
repoId: string;
taskId: string;
repoRemote: string;
repoLocalPath: string;
repoLocalPath?: string;
branchName: string | null;
title: string | null;
task: string;

30
foundry/packages/backend/src/actors/task/workbench.ts
View file

@ -7,7 +7,6 @@ import { getOrCreateTaskStatusSync, getOrCreateProject, getOrCreateWorkspace, ge
import { resolveWorkspaceGithubAuth } from "../../services/github-auth.js";
import { task as taskTable, taskRuntime, taskWorkbenchSessions } from "./db/schema.js";
import { getCurrentRecord } from "./workflow/common.js";
import { taskWorkflowQueueName } from "./workflow/queue.js";
const STATUS_SYNC_INTERVAL_MS = 1_000;
@ -599,7 +598,13 @@ export async function ensureWorkbenchSeeded(c: any): Promise<any> {
function buildSessionSummary(record: any, meta: any): any {
const derivedSandboxSessionId = meta.sandboxSessionId ?? (meta.status === "pending_provision" && record.activeSessionId ? record.activeSessionId : null);
const sessionStatus =
meta.status === "ready" && derivedSandboxSessionId ? activeSessionStatus(record, derivedSandboxSessionId) : meta.status === "error" ? "error" : "idle";
meta.status === "pending_provision" || meta.status === "pending_session_create"
? meta.status
: meta.status === "ready" && derivedSandboxSessionId
? activeSessionStatus(record, derivedSandboxSessionId)
: meta.status === "error"
? "error"
: "ready";
let thinkingSinceMs = meta.thinkingSinceMs ?? null;
let unread = Boolean(meta.unread);
if (thinkingSinceMs && sessionStatus !== "running") {
@ -617,6 +622,7 @@ function buildSessionSummary(record: any, meta: any): any {
thinkingSinceMs: sessionStatus === "running" ? thinkingSinceMs : null,
unread,
created: Boolean(meta.created || derivedSandboxSessionId),
errorMessage: meta.errorMessage ?? null,
};
}
@ -633,6 +639,7 @@ function buildSessionDetailFromMeta(record: any, meta: any): any {
thinkingSinceMs: summary.thinkingSinceMs,
unread: summary.unread,
created: summary.created,
errorMessage: summary.errorMessage,
draft: {
text: meta.draftText ?? "",
attachments: Array.isArray(meta.draftAttachments) ? meta.draftAttachments : [],
@ -655,7 +662,7 @@ export async function buildTaskSummary(c: any): Promise<any> {
id: c.state.taskId,
repoId: c.state.repoId,
title: record.title ?? "New Task",
status: record.status === "archived" ? "archived" : record.status === "running" ? "running" : record.status === "idle" ? "idle" : "new",
status: record.status ?? "new",
repoName: repoLabelFromRemote(c.state.repoRemote),
updatedAtMs: record.updatedAt,
branch: record.branchName,
@ -837,14 +844,6 @@ export async function renameWorkbenchBranch(c: any, value: string): Promise<void
export async function createWorkbenchSession(c: any, model?: string): Promise<{ tabId: string }> {
let record = await ensureWorkbenchSeeded(c);
if (!record.activeSandboxId) {
// Fire-and-forget: enqueue provisioning without waiting to avoid self-deadlock
// (this handler already runs inside the task workflow loop, so wait:true would deadlock).
const providerId = record.providerId ?? c.state.providerId ?? getActorRuntimeContext().providers.defaultProviderId();
await selfTask(c).send(taskWorkflowQueueName("task.command.provision"), { providerId }, { wait: false });
throw new Error("sandbox is provisioning — retry shortly");
}
if (record.activeSessionId) {
const existingSessions = await listSessionMetaRows(c);
if (existingSessions.length === 0) {
@ -1216,9 +1215,16 @@ export async function publishWorkbenchPr(c: any): Promise<void> {
if (!record.branchName) {
throw new Error("cannot publish PR without a branch");
}
let repoLocalPath = c.state.repoLocalPath;
if (!repoLocalPath) {
const project = await getOrCreateProject(c, c.state.workspaceId, c.state.repoId, c.state.repoRemote);
const result = await project.ensure({ remoteUrl: c.state.repoRemote });
repoLocalPath = result.localPath;
c.state.repoLocalPath = repoLocalPath;
}
const { driver } = getActorRuntimeContext();
const auth = await resolveWorkspaceGithubAuth(c, c.state.workspaceId);
const created = await driver.github.createPr(c.state.repoLocalPath, record.branchName, record.title ?? c.state.task, undefined, {
const created = await driver.github.createPr(repoLocalPath, record.branchName, record.title ?? c.state.task, undefined, {
githubToken: auth?.githubToken ?? null,
});
await c.db

3
foundry/packages/backend/src/actors/task/workflow/index.ts
View file

@ -76,6 +76,7 @@ const commandHandlers: Record<TaskQueueName, WorkflowHandler> = {
"task.command.provision": async (loopCtx, msg) => {
const body = msg.body;
await loopCtx.removed("init-failed", "step");
await loopCtx.removed("init-failed-v2", "step");
try {
await loopCtx.step("init-ensure-name", async () => initEnsureNameActivity(loopCtx));
await loopCtx.step("init-assert-name", async () => initAssertNameActivity(loopCtx));
@ -107,7 +108,7 @@ const commandHandlers: Record<TaskQueueName, WorkflowHandler> = {
await loopCtx.step("init-complete", async () => initCompleteActivity(loopCtx, body, sandbox, session));
await msg.complete({ ok: true });
} catch (error) {
await loopCtx.step("init-failed-v2", async () => initFailedActivity(loopCtx, error));
await loopCtx.step("init-failed-v3", async () => initFailedActivity(loopCtx, error));
await msg.complete({
ok: false,
error: resolveErrorMessage(error),

12
foundry/packages/backend/src/actors/task/workflow/init.ts
View file

@ -178,8 +178,16 @@ export async function initEnsureNameActivity(loopCtx: any): Promise<void> {
const { driver } = getActorRuntimeContext();
const auth = await resolveWorkspaceGithubAuth(loopCtx, loopCtx.state.workspaceId);
let repoLocalPath = loopCtx.state.repoLocalPath;
if (!repoLocalPath) {
const project = await getOrCreateProject(loopCtx, loopCtx.state.workspaceId, loopCtx.state.repoId, loopCtx.state.repoRemote);
const result = await project.ensure({ remoteUrl: loopCtx.state.repoRemote });
repoLocalPath = result.localPath;
loopCtx.state.repoLocalPath = repoLocalPath;
}
try {
await driver.git.fetch(loopCtx.state.repoLocalPath, { githubToken: auth?.githubToken ?? null });
await driver.git.fetch(repoLocalPath, { githubToken: auth?.githubToken ?? null });
} catch (error) {
logActorWarning("task.init", "fetch before naming failed", {
workspaceId: loopCtx.state.workspaceId,
@ -188,7 +196,7 @@ export async function initEnsureNameActivity(loopCtx: any): Promise<void> {
error: resolveErrorMessage(error),
});
}
const remoteBranches = (await driver.git.listRemoteBranches(loopCtx.state.repoLocalPath, { githubToken: auth?.githubToken ?? null })).map(
const remoteBranches = (await driver.git.listRemoteBranches(repoLocalPath, { githubToken: auth?.githubToken ?? null })).map(
(branch: any) => branch.branchName,
);

5
foundry/packages/backend/src/actors/workspace/actions.ts
View file

@ -370,7 +370,6 @@ async function createTaskMutation(c: any, input: CreateTaskInput): Promise<TaskR
.run();
const project = await getOrCreateProject(c, c.state.workspaceId, repoId, remoteUrl);
await project.ensure({ remoteUrl });
const created = await project.createTask({
task: input.task,
@ -457,7 +456,7 @@ export async function runWorkspaceWorkflow(ctx: any): Promise<void> {
if (msg.name === "workspace.command.createTask") {
const result = await loopCtx.step({
name: "workspace-create-task",
timeout: 12 * 60_000,
timeout: 60_000,
run: async () => createTaskMutation(loopCtx, msg.body as CreateTaskInput),
});
await msg.complete(result);
@ -547,7 +546,7 @@ export const workspaceActions = {
return expectQueueResponse<TaskRecord>(
await self.send(workspaceWorkflowQueueName("workspace.command.createTask"), input, {
wait: true,
timeout: 12 * 60_000,
timeout: 60_000,
}),
);
},

95
foundry/packages/backend/src/providers/daytona/index.ts
View file

@ -17,6 +17,9 @@ import type {
} from "../provider-api/index.js";
import type { DaytonaDriver } from "../../driver.js";
import { Image } from "@daytonaio/sdk";
import { readFileSync } from "node:fs";
import { homedir } from "node:os";
import { resolve } from "node:path";
export interface DaytonaProviderConfig {
endpoint?: string;
@ -176,6 +179,51 @@ export class DaytonaProvider implements SandboxProvider {
}
}
private shellSingleQuote(value: string): string {
return `'${value.replace(/'/g, `'\"'\"'`)}'`;
}
private readLocalCodexAuth(): string | null {
const authPath = resolve(homedir(), ".codex", "auth.json");
try {
return readFileSync(authPath, "utf8");
} catch {
return null;
}
}
private buildCloneRepoScript(req: CreateSandboxRequest, repoDir: string): string {
const usesGithubHttpAuth = req.repoRemote.startsWith("https://github.com/");
const githubPath = usesGithubHttpAuth ? req.repoRemote.slice("https://github.com/".length) : "";
const lines = [
"set -eu",
"export GIT_TERMINAL_PROMPT=0",
"export GIT_ASKPASS=/bin/echo",
`TOKEN=${JSON.stringify(req.githubToken ?? "")}`,
'if [ -z "$TOKEN" ]; then',
' if [ -n "${GH_TOKEN:-}" ]; then TOKEN="$GH_TOKEN"; else TOKEN="${GITHUB_TOKEN:-}"; fi',
"fi",
'AUTH_REMOTE=""',
...(usesGithubHttpAuth ? ['if [ -n "$TOKEN" ]; then', ` AUTH_REMOTE="https://x-access-token:${"$"}TOKEN@github.com/${githubPath}"`, "fi"] : []),
`rm -rf "${repoDir}"`,
`mkdir -p "${repoDir}"`,
`rmdir "${repoDir}"`,
// Foundry test repos can be private, so clone/fetch must use the sandbox's GitHub token when available.
...(usesGithubHttpAuth
? ['if [ -n "$AUTH_REMOTE" ]; then', ` git clone "$AUTH_REMOTE" "${repoDir}"`, "else", ` git clone "${req.repoRemote}" "${repoDir}"`, "fi"]
: [`git clone "${req.repoRemote}" "${repoDir}"`]),
`cd "${repoDir}"`,
...(usesGithubHttpAuth ? ['if [ -n "$AUTH_REMOTE" ]; then', ` git remote set-url origin "${req.repoRemote}"`, "fi"] : []),
// The task branch may not exist remotely yet (agent push creates it). Base off current branch (default branch).
`if git show-ref --verify --quiet "refs/remotes/origin/${req.branchName}"; then git checkout -B "${req.branchName}" "origin/${req.branchName}"; else git checkout -B "${req.branchName}" "$(git branch --show-current 2>/dev/null || echo main)"; fi`,
`git config user.email "foundry@local" >/dev/null 2>&1 || true`,
`git config user.name "Foundry" >/dev/null 2>&1 || true`,
];
return lines.join("\n");
}
id() {
return "daytona" as const;
}
@ -242,37 +290,7 @@ export class DaytonaProvider implements SandboxProvider {
});
const cloneStartedAt = Date.now();
await this.runCheckedCommand(
sandbox.id,
[
"bash",
"-lc",
`${JSON.stringify(
[
"set -euo pipefail",
"export GIT_TERMINAL_PROMPT=0",
"export GIT_ASKPASS=/bin/echo",
`TOKEN=${JSON.stringify(req.githubToken ?? "")}`,
'if [ -z "$TOKEN" ]; then TOKEN="${GH_TOKEN:-${GITHUB_TOKEN:-}}"; fi',
"GIT_AUTH_ARGS=()",
`if [ -n "$TOKEN" ] && [[ "${req.repoRemote}" == https://github.com/* ]]; then AUTH_HEADER="$(printf 'x-access-token:%s' "$TOKEN" | base64 | tr -d '\\n')"; GIT_AUTH_ARGS=(-c "http.https://github.com/.extraheader=AUTHORIZATION: basic $AUTH_HEADER"); fi`,
`rm -rf "${repoDir}"`,
`mkdir -p "${repoDir}"`,
`rmdir "${repoDir}"`,
// Foundry test repos can be private, so clone/fetch must use the sandbox's GitHub token when available.
`git "\${GIT_AUTH_ARGS[@]}" clone "${req.repoRemote}" "${repoDir}"`,
`cd "${repoDir}"`,
`if [ -n "$TOKEN" ] && [[ "${req.repoRemote}" == https://github.com/* ]]; then git config --local credential.helper ""; git config --local http.https://github.com/.extraheader "AUTHORIZATION: basic $AUTH_HEADER"; fi`,
`git "\${GIT_AUTH_ARGS[@]}" fetch origin --prune`,
// The task branch may not exist remotely yet (agent push creates it). Base off current branch (default branch).
`if git show-ref --verify --quiet "refs/remotes/origin/${req.branchName}"; then git checkout -B "${req.branchName}" "origin/${req.branchName}"; else git checkout -B "${req.branchName}" "$(git branch --show-current 2>/dev/null || echo main)"; fi`,
`git config user.email "foundry@local" >/dev/null 2>&1 || true`,
`git config user.name "Foundry" >/dev/null 2>&1 || true`,
].join("; "),
)}`,
].join(" "),
"clone repo",
);
await this.runCheckedCommand(sandbox.id, ["bash", "-lc", this.shellSingleQuote(this.buildCloneRepoScript(req, repoDir))].join(" "), "clone repo");
emitDebug("daytona.createSandbox.clone_repo.done", {
sandboxId: sandbox.id,
durationMs: Date.now() - cloneStartedAt,
@ -357,6 +375,15 @@ export class DaytonaProvider implements SandboxProvider {
const sandboxAgentExports = this.buildShellExports({
SANDBOX_AGENT_ACP_REQUEST_TIMEOUT_MS: acpRequestTimeoutMs.toString(),
});
const codexAuth = this.readLocalCodexAuth();
const codexAuthSetup = codexAuth
? [
'mkdir -p "$HOME/.codex" "$HOME/.config/codex"',
`printf %s ${JSON.stringify(Buffer.from(codexAuth, "utf8").toString("base64"))} | base64 -d > "$HOME/.codex/auth.json"`,
'cp "$HOME/.codex/auth.json" "$HOME/.config/codex/auth.json"',
"unset OPENAI_API_KEY CODEX_API_KEY",
]
: [];
await this.ensureStarted(req.sandboxId);
@ -407,16 +434,16 @@ export class DaytonaProvider implements SandboxProvider {
[
"bash",
"-lc",
JSON.stringify(
this.shellSingleQuote(
[
"set -euo pipefail",
'export PATH="$HOME/.local/bin:$PATH"',
...sandboxAgentExports,
...codexAuthSetup,
"command -v sandbox-agent >/dev/null 2>&1",
"if pgrep -x sandbox-agent >/dev/null; then exit 0; fi",
'rm -f "$HOME/.codex/auth.json" "$HOME/.config/codex/auth.json"',
`nohup sandbox-agent server --no-token --host 0.0.0.0 --port ${DaytonaProvider.SANDBOX_AGENT_PORT} >/tmp/sandbox-agent.log 2>&1 &`,
].join("; "),
].join("\n"),
),
].join(" "),
"start sandbox-agent",

27
foundry/packages/backend/test/daytona-provider.test.ts
View file

@ -1,3 +1,6 @@
import { mkdirSync, rmSync, writeFileSync } from "node:fs";
import { tmpdir } from "node:os";
import { resolve } from "node:path";
import { describe, expect, it } from "vitest";
import type { DaytonaClientLike, DaytonaDriver } from "../src/driver.js";
import type { DaytonaCreateSandboxOptions } from "../src/integrations/daytona/client.js";
@ -91,6 +94,10 @@ describe("daytona provider snapshot image behavior", () => {
const commands = client.executedCommands.join("\n");
expect(commands).toContain("GIT_TERMINAL_PROMPT=0");
expect(commands).toContain("GIT_ASKPASS=/bin/echo");
expect(commands).not.toContain("[[");
expect(commands).not.toContain("GIT_AUTH_ARGS=()");
expect(commands).not.toContain("${GIT_AUTH_ARGS[@]}");
expect(commands).not.toContain(".extraheader");
expect(handle.metadata.snapshot).toBe("snapshot-foundry");
expect(handle.metadata.image).toBe("ubuntu:24.04");
@ -100,6 +107,11 @@ describe("daytona provider snapshot image behavior", () => {
it("starts sandbox-agent with ACP timeout env override", async () => {
const previous = process.env.HF_SANDBOX_AGENT_ACP_REQUEST_TIMEOUT_MS;
const previousHome = process.env.HOME;
const tempHome = resolve(tmpdir(), `daytona-provider-test-${Date.now()}`);
mkdirSync(resolve(tempHome, ".codex"), { recursive: true });
writeFileSync(resolve(tempHome, ".codex", "auth.json"), JSON.stringify({ access_token: "test-token" }));
process.env.HOME = tempHome;
process.env.HF_SANDBOX_AGENT_ACP_REQUEST_TIMEOUT_MS = "240000";
try {
@ -111,15 +123,18 @@ describe("daytona provider snapshot image behavior", () => {
sandboxId: "sandbox-1",
});
const startCommand = client.executedCommands.find((command) =>
command.includes("nohup env SANDBOX_AGENT_ACP_REQUEST_TIMEOUT_MS=240000 sandbox-agent server"),
const startCommand = client.executedCommands.find(
(command) => command.includes("export SANDBOX_AGENT_ACP_REQUEST_TIMEOUT_MS=") && command.includes("sandbox-agent server --no-token"),
);
const joined = client.executedCommands.join("\n");
expect(joined).toContain("sandbox-agent/0.3.0/install.sh");
expect(joined).toContain("SANDBOX_AGENT_ACP_REQUEST_TIMEOUT_MS=240000");
expect(joined).toContain("SANDBOX_AGENT_ACP_REQUEST_TIMEOUT_MS");
expect(joined).toContain("apt-get install -y nodejs npm");
expect(joined).toContain("sandbox-agent server --no-token --host 0.0.0.0 --port 2468");
expect(joined).toContain('mkdir -p "$HOME/.codex" "$HOME/.config/codex"');
expect(joined).toContain("unset OPENAI_API_KEY CODEX_API_KEY");
expect(joined).not.toContain('rm -f "$HOME/.codex/auth.json"');
expect(startCommand).toBeTruthy();
} finally {
if (previous === undefined) {
@ -127,6 +142,12 @@ describe("daytona provider snapshot image behavior", () => {
} else {
process.env.HF_SANDBOX_AGENT_ACP_REQUEST_TIMEOUT_MS = previous;
}
if (previousHome === undefined) {
delete process.env.HOME;
} else {
process.env.HOME = previousHome;
}
rmSync(tempHome, { force: true, recursive: true });
}
});

2
foundry/packages/client/src/backend-client.ts
View file

@ -54,7 +54,7 @@ export interface SandboxSessionRecord {
lastConnectionId: string;
createdAt: number;
destroyedAt?: number;
status?: "running" | "idle" | "error";
status?: "pending_provision" | "pending_session_create" | "ready" | "running" | "idle" | "error";
}
export interface SandboxSessionEventRecord {

9
foundry/packages/client/src/interest/manager.ts
View file

@ -2,6 +2,14 @@ import type { TopicData, TopicKey, TopicParams } from "./topics.js";
export type TopicStatus = "loading" | "connected" | "error";
export interface DebugInterestTopic {
topicKey: TopicKey;
cacheKey: string;
listenerCount: number;
status: TopicStatus;
lastRefreshAt: number | null;
}
export interface TopicState<K extends TopicKey> {
data: TopicData<K> | undefined;
status: TopicStatus;
@ -20,5 +28,6 @@ export interface InterestManager {
getSnapshot<K extends TopicKey>(topicKey: K, params: TopicParams<K>): TopicData<K> | undefined;
getStatus<K extends TopicKey>(topicKey: K, params: TopicParams<K>): TopicStatus;
getError<K extends TopicKey>(topicKey: K, params: TopicParams<K>): Error | null;
listDebugTopics(): DebugInterestTopic[];
dispose(): void;
}

27
foundry/packages/client/src/interest/remote-manager.ts
View file

@ -1,5 +1,5 @@
import type { BackendClient } from "../backend-client.js";
import type { InterestManager, TopicStatus } from "./manager.js";
import type { DebugInterestTopic, InterestManager, TopicStatus } from "./manager.js";
import { topicDefinitions, type TopicData, type TopicDefinition, type TopicKey, type TopicParams } from "./topics.js";
const GRACE_PERIOD_MS = 30_000;
@ -19,7 +19,7 @@ export class RemoteInterestManager implements InterestManager {
let entry = this.entries.get(cacheKey);
if (!entry) {
entry = new TopicEntry(definition, this.backend, params as any);
entry = new TopicEntry(topicKey, cacheKey, definition, this.backend, params as any);
this.entries.set(cacheKey, entry);
}
@ -53,6 +53,13 @@ export class RemoteInterestManager implements InterestManager {
return this.entries.get((topicDefinitions[topicKey] as any).key(params))?.error ?? null;
}
listDebugTopics(): DebugInterestTopic[] {
return [...this.entries.values()]
.filter((entry) => entry.listenerCount > 0)
.map((entry) => entry.getDebugTopic())
.sort((left, right) => left.cacheKey.localeCompare(right.cacheKey));
}
dispose(): void {
for (const entry of this.entries.values()) {
entry.dispose();
@ -66,6 +73,7 @@ class TopicEntry<TData, TParams, TEvent> {
status: TopicStatus = "loading";
error: Error | null = null;
listenerCount = 0;
lastRefreshAt: number | null = null;
private readonly listeners = new Set<() => void>();
private conn: Awaited<ReturnType<TopicDefinition<TData, TParams, TEvent>["connect"]>> | null = null;
@ -76,11 +84,23 @@ class TopicEntry<TData, TParams, TEvent> {
private started = false;
constructor(
private readonly topicKey: TopicKey,
private readonly cacheKey: string,
private readonly definition: TopicDefinition<TData, TParams, TEvent>,
private readonly backend: BackendClient,
private readonly params: TParams,
) {}
getDebugTopic(): DebugInterestTopic {
return {
topicKey: this.topicKey,
cacheKey: this.cacheKey,
listenerCount: this.listenerCount,
status: this.status,
lastRefreshAt: this.lastRefreshAt,
};
}
addListener(listener: () => void): void {
this.listeners.add(listener);
this.listenerCount = this.listeners.size;
@ -125,6 +145,7 @@ class TopicEntry<TData, TParams, TEvent> {
this.data = undefined;
this.status = "loading";
this.error = null;
this.lastRefreshAt = null;
this.started = false;
}
@ -140,6 +161,7 @@ class TopicEntry<TData, TParams, TEvent> {
return;
}
this.data = this.definition.applyEvent(this.data, event);
this.lastRefreshAt = Date.now();
this.notify();
});
this.unsubscribeError = this.conn.onError((error: unknown) => {
@ -149,6 +171,7 @@ class TopicEntry<TData, TParams, TEvent> {
});
this.data = await this.definition.fetchInitial(this.backend, this.params);
this.status = "connected";
this.lastRefreshAt = Date.now();
this.started = true;
this.notify();
} catch (error) {

11
foundry/packages/client/test/interest-manager.test.ts
View file

@ -104,6 +104,14 @@ describe("RemoteInterestManager", () => {
expect(backend.getWorkspaceSummary).toHaveBeenCalledTimes(1);
expect(manager.getStatus("workspace", params)).toBe("connected");
expect(manager.getSnapshot("workspace", params)?.taskSummaries[0]?.title).toBe("Initial task");
expect(manager.listDebugTopics()).toEqual([
expect.objectContaining({
topicKey: "workspace",
cacheKey: "workspace:ws-1",
listenerCount: 2,
status: "connected",
}),
]);
conn.emit("workspaceUpdated", {
type: "taskSummaryUpdated",
@ -123,6 +131,7 @@ describe("RemoteInterestManager", () => {
expect(manager.getSnapshot("workspace", params)?.taskSummaries[0]?.title).toBe("Updated task");
expect(listenerA).toHaveBeenCalled();
expect(listenerB).toHaveBeenCalled();
expect(manager.listDebugTopics()[0]?.lastRefreshAt).toEqual(expect.any(Number));
unsubscribeA();
unsubscribeB();
@ -140,6 +149,7 @@ describe("RemoteInterestManager", () => {
unsubscribeA();
vi.advanceTimersByTime(29_000);
expect(manager.listDebugTopics()).toEqual([]);
const unsubscribeB = manager.subscribe("workspace", params, () => {});
await flushAsyncWork();
@ -148,6 +158,7 @@ describe("RemoteInterestManager", () => {
expect(conn.disposeCount).toBe(0);
unsubscribeB();
expect(manager.listDebugTopics()).toEqual([]);
vi.advanceTimersByTime(30_000);
expect(conn.disposeCount).toBe(1);

58
foundry/packages/frontend/src/components/dev-panel.tsx
View file

@ -2,7 +2,9 @@ import { memo, useCallback, useEffect, useMemo, useState } from "react";
import { useStyletron } from "baseui";
import { useFoundryTokens } from "../app/theme";
import { isMockFrontendClient } from "../lib/env";
import { interestManager } from "../lib/interest";
import type { FoundryOrganization, TaskWorkbenchSnapshot, WorkbenchTask } from "@sandbox-agent/foundry-shared";
import type { DebugInterestTopic } from "@sandbox-agent/foundry-client";
interface DevPanelProps {
workspaceId: string;
@ -15,9 +17,25 @@ interface TopicInfo {
key: string;
listenerCount: number;
hasConnection: boolean;
status: "loading" | "connected" | "error";
lastRefresh: number | null;
}
function topicLabel(topic: DebugInterestTopic): string {
switch (topic.topicKey) {
case "app":
return "App";
case "workspace":
return "Workspace";
case "task":
return "Task";
case "session":
return "Session";
case "sandboxProcesses":
return "Sandbox";
}
}
function timeAgo(ts: number | null): string {
if (!ts) return "never";
const seconds = Math.floor((Date.now() - ts) / 1000);
@ -37,8 +55,11 @@ function taskStatusLabel(task: WorkbenchTask): string {
function statusColor(status: string, t: ReturnType<typeof useFoundryTokens>): string {
switch (status) {
case "connected":
case "running":
return t.statusSuccess;
case "loading":
return t.statusWarning;
case "archived":
return t.textMuted;
case "error":
@ -88,33 +109,15 @@ export const DevPanel = memo(function DevPanel({ workspaceId, snapshot, organiza
}, []);
const topics = useMemo((): TopicInfo[] => {
const items: TopicInfo[] = [];
// Workbench subscription topic
items.push({
label: "Workbench",
key: `ws:${workspaceId}`,
listenerCount: 1,
hasConnection: true,
lastRefresh: now,
});
// Per-task tab subscriptions
for (const task of snapshot.tasks ?? []) {
if (task.status === "archived") continue;
for (const tab of task.tabs ?? []) {
items.push({
label: `Tab/${task.title?.slice(0, 16) || task.id.slice(0, 8)}/${tab.sessionName.slice(0, 10)}`,
key: `${workspaceId}:${task.id}:${tab.id}`,
listenerCount: 1,
hasConnection: tab.status === "running",
lastRefresh: tab.status === "running" ? now : null,
});
}
}
return items;
}, [workspaceId, snapshot, now]);
return interestManager.listDebugTopics().map((topic) => ({
label: topicLabel(topic),
key: topic.cacheKey,
listenerCount: topic.listenerCount,
hasConnection: topic.status === "connected",
status: topic.status,
lastRefresh: topic.lastRefreshAt,
}));
}, [now]);
const tasks = snapshot.tasks ?? [];
const repos = snapshot.repos ?? [];
@ -199,6 +202,7 @@ export const DevPanel = memo(function DevPanel({ workspaceId, snapshot, organiza
<span className={css({ fontSize: "10px", color: t.textPrimary, flex: 1, overflow: "hidden", textOverflow: "ellipsis", whiteSpace: "nowrap" })}>
{topic.label}
</span>
<span className={`${mono} ${css({ color: statusColor(topic.status, t) })}`}>{topic.status}</span>
<span className={`${mono} ${css({ color: t.textMuted })}`}>{topic.key.length > 24 ? `...${topic.key.slice(-20)}` : topic.key}</span>
<span className={`${mono} ${css({ color: t.textTertiary })}`}>{timeAgo(topic.lastRefresh)}</span>
</div>

136
foundry/packages/frontend/src/components/mock-layout.tsx
View file

@ -22,7 +22,7 @@ import { Sidebar } from "./mock-layout/sidebar";
import { TabStrip } from "./mock-layout/tab-strip";
import { TerminalPane } from "./mock-layout/terminal-pane";
import { TranscriptHeader } from "./mock-layout/transcript-header";
import { PROMPT_TEXTAREA_MAX_HEIGHT, PROMPT_TEXTAREA_MIN_HEIGHT, SPanel, ScrollBody, Shell } from "./mock-layout/ui";
import { PROMPT_TEXTAREA_MAX_HEIGHT, PROMPT_TEXTAREA_MIN_HEIGHT, SPanel, ScrollBody, Shell, SpinnerDot } from "./mock-layout/ui";
import { DevPanel, useDevPanel } from "./dev-panel";
import {
buildDisplayMessages,
@ -88,6 +88,7 @@ function toLegacyTab(
thinkingSinceMs: summary.thinkingSinceMs,
unread: summary.unread,
created: summary.created,
errorMessage: summary.errorMessage ?? null,
draft: sessionDetail?.draft ?? {
text: "",
attachments: [],
@ -107,7 +108,9 @@ function toLegacyTask(
id: summary.id,
repoId: summary.repoId,
title: detail?.title ?? summary.title,
status: detail?.status ?? summary.status,
status: detail?.runtimeStatus ?? detail?.status ?? summary.status,
runtimeStatus: detail?.runtimeStatus,
statusMessage: detail?.statusMessage ?? null,
repoName: detail?.repoName ?? summary.repoName,
updatedAtMs: detail?.updatedAtMs ?? summary.updatedAtMs,
branch: detail?.branch ?? summary.branch,
@ -117,9 +120,30 @@ function toLegacyTask(
diffs: detail?.diffs ?? {},
fileTree: detail?.fileTree ?? [],
minutesUsed: detail?.minutesUsed ?? 0,
activeSandboxId: detail?.activeSandboxId ?? null,
};
}
function isProvisioningTaskStatus(status: string | null | undefined): boolean {
return status === "new" || String(status ?? "").startsWith("init_");
}
function sessionStateMessage(tab: Task["tabs"][number] | null | undefined): string | null {
if (!tab) {
return null;
}
if (tab.status === "pending_provision") {
return "Provisioning sandbox...";
}
if (tab.status === "pending_session_create") {
return "Creating session...";
}
if (tab.status === "error") {
return tab.errorMessage ?? "Session failed to start.";
}
return null;
}
function groupProjects(repos: Array<{ id: string; label: string }>, tasks: Task[]) {
return repos
.map((repo) => ({
@ -202,6 +226,14 @@ const TranscriptPanel = memo(function TranscriptPanel({
const isTerminal = task.status === "archived";
const historyEvents = useMemo(() => buildHistoryEvents(task.tabs), [task.tabs]);
const activeMessages = useMemo(() => buildDisplayMessages(activeAgentTab), [activeAgentTab]);
const taskProvisioning = isProvisioningTaskStatus(task.runtimeStatus ?? task.status);
const taskProvisioningMessage = task.statusMessage ?? "Provisioning sandbox...";
const activeSessionMessage = sessionStateMessage(activeAgentTab);
const showPendingSessionState =
!activeDiff &&
!!activeAgentTab &&
(activeAgentTab.status === "pending_provision" || activeAgentTab.status === "pending_session_create" || activeAgentTab.status === "error") &&
activeMessages.length === 0;
const draft = promptTab?.draft.text ?? "";
const attachments = promptTab?.draft.attachments ?? [];
@ -619,26 +651,88 @@ const TranscriptPanel = memo(function TranscriptPanel({
display: "flex",
flexDirection: "column",
gap: "12px",
alignItems: "center",
}}
>
<h2 style={{ margin: 0, fontSize: "20px", fontWeight: 600 }}>Create the first session</h2>
<p style={{ margin: 0, opacity: 0.75 }}>Sessions are where you chat with the agent. Start one now to send the first prompt on this task.</p>
<button
type="button"
onClick={addTab}
style={{
alignSelf: "center",
border: 0,
borderRadius: "999px",
padding: "10px 18px",
background: t.borderMedium,
color: t.textPrimary,
cursor: "pointer",
fontWeight: 600,
}}
>
New session
</button>
{taskProvisioning ? (
<>
<SpinnerDot size={16} />
<h2 style={{ margin: 0, fontSize: "20px", fontWeight: 600 }}>Provisioning task</h2>
<p style={{ margin: 0, opacity: 0.75 }}>{taskProvisioningMessage}</p>
</>
) : (
<>
<h2 style={{ margin: 0, fontSize: "20px", fontWeight: 600 }}>Create the first session</h2>
<p style={{ margin: 0, opacity: 0.75 }}>Sessions are where you chat with the agent. Start one now to send the first prompt on this task.</p>
<button
type="button"
onClick={addTab}
style={{
alignSelf: "center",
border: 0,
borderRadius: "999px",
padding: "10px 18px",
background: t.borderMedium,
color: t.textPrimary,
cursor: "pointer",
fontWeight: 600,
}}
>
New session
</button>
</>
)}
</div>
</div>
</ScrollBody>
) : showPendingSessionState ? (
<ScrollBody>
<div
style={{
minHeight: "100%",
display: "flex",
alignItems: "center",
justifyContent: "center",
padding: "32px",
}}
>
<div
style={{
maxWidth: "420px",
textAlign: "center",
display: "flex",
flexDirection: "column",
gap: "12px",
alignItems: "center",
}}
>
{activeAgentTab?.status === "error" ? null : <SpinnerDot size={16} />}
<h2 style={{ margin: 0, fontSize: "20px", fontWeight: 600 }}>
{activeAgentTab?.status === "pending_provision"
? "Provisioning sandbox"
: activeAgentTab?.status === "pending_session_create"
? "Creating session"
: "Session unavailable"}
</h2>
<p style={{ margin: 0, opacity: 0.75 }}>{activeSessionMessage}</p>
{activeAgentTab?.status === "error" ? (
<button
type="button"
onClick={addTab}
style={{
alignSelf: "center",
border: 0,
borderRadius: "999px",
padding: "10px 18px",
background: t.borderMedium,
color: t.textPrimary,
cursor: "pointer",
fontWeight: 600,
}}
>
Retry session
</button>
) : null}
</div>
</div>
</ScrollBody>
@ -658,7 +752,7 @@ const TranscriptPanel = memo(function TranscriptPanel({
/>
</ScrollBody>
)}
{!isTerminal && promptTab ? (
{!isTerminal && promptTab && (promptTab.status === "ready" || promptTab.status === "running" || promptTab.status === "idle") ? (
<PromptComposer
draft={draft}
textareaRef={textareaRef}

6
foundry/packages/frontend/src/components/mock-layout/sidebar.tsx
View file

@ -521,6 +521,10 @@ export const Sidebar = memo(function Sidebar({
const isActive = task.id === activeId;
const isDim = task.status === "archived";
const isRunning = task.tabs.some((tab) => tab.status === "running");
const isProvisioning =
String(task.status).startsWith("init_") ||
task.status === "new" ||
task.tabs.some((tab) => tab.status === "pending_provision" || tab.status === "pending_session_create");
const hasUnread = task.tabs.some((tab) => tab.unread);
const isDraft = task.pullRequest == null || task.pullRequest.status === "draft";
const totalAdded = task.fileChanges.reduce((sum, file) => sum + file.added, 0);
@ -592,7 +596,7 @@ export const Sidebar = memo(function Sidebar({
flexShrink: 0,
})}
>
<TaskIndicator isRunning={isRunning} hasUnread={hasUnread} isDraft={isDraft} />
<TaskIndicator isRunning={isRunning} isProvisioning={isProvisioning} hasUnread={hasUnread} isDraft={isDraft} />
</div>
<LabelSmall
$style={{

15
foundry/packages/frontend/src/components/mock-layout/ui.tsx
View file

@ -118,10 +118,21 @@ export const UnreadDot = memo(function UnreadDot() {
);
});
export const TaskIndicator = memo(function TaskIndicator({ isRunning, hasUnread, isDraft }: { isRunning: boolean; hasUnread: boolean; isDraft: boolean }) {
export const TaskIndicator = memo(function TaskIndicator({
isRunning,
isProvisioning,
hasUnread,
isDraft,
}: {
isRunning: boolean;
isProvisioning: boolean;
hasUnread: boolean;
isDraft: boolean;
}) {
const t = useFoundryTokens();
if (isRunning) return <SpinnerDot size={8} />;
if (isProvisioning) return <SpinnerDot size={8} />;
if (hasUnread) return <UnreadDot />;
if (isDraft) return <GitPullRequestDraft size={12} color={t.textSecondary} />;
return <GitPullRequest size={12} color={t.statusSuccess} />;
@ -174,7 +185,7 @@ export const AgentIcon = memo(function AgentIcon({ agent, size = 14 }: { agent:
});
export const TabAvatar = memo(function TabAvatar({ tab }: { tab: AgentTab }) {
if (tab.status === "running") return <SpinnerDot size={8} />;
if (tab.status === "running" || tab.status === "pending_provision" || tab.status === "pending_session_create") return <SpinnerDot size={8} />;
if (tab.unread) return <UnreadDot />;
return <AgentIcon agent={tab.agent} size={13} />;
});

66
foundry/packages/frontend/src/components/workspace-dashboard.tsx
View file

@ -99,7 +99,8 @@ const AGENT_OPTIONS: SelectItem[] = [
function statusKind(status: WorkbenchTaskStatus): StatusTagKind {
if (status === "running") return "positive";
if (status === "new") return "warning";
if (status === "error") return "negative";
if (status === "new" || String(status).startsWith("init_")) return "warning";
return "neutral";
}
@ -497,6 +498,10 @@ export function WorkspaceDashboard({ workspaceId, selectedTaskId, selectedRepoId
}
: null,
);
const selectedSessionSummary = useMemo(() => sessionRows.find((session) => session.id === resolvedSessionId) ?? null, [resolvedSessionId, sessionRows]);
const isPendingProvision = selectedSessionSummary?.status === "pending_provision";
const isPendingSessionCreate = selectedSessionSummary?.status === "pending_session_create";
const isSessionError = selectedSessionSummary?.status === "error";
const canStartSession = Boolean(selectedForSession && activeSandbox?.sandboxId);
const startSessionFromTask = async (): Promise<{ id: string; status: "running" | "idle" | "error" }> => {
@ -1363,19 +1368,47 @@ export function WorkspaceDashboard({ workspaceId, selectedTaskId, selectedRepoId
>
{resolvedSessionId && sessionState.status === "loading" ? <Skeleton rows={2} height="90px" /> : null}
{selectedSessionSummary && (isPendingProvision || isPendingSessionCreate) ? (
<div
className={css({
display: "flex",
flexDirection: "column",
gap: theme.sizing.scale300,
padding: theme.sizing.scale500,
border: `1px solid ${theme.colors.borderOpaque}`,
backgroundColor: theme.colors.backgroundSecondary,
marginBottom: theme.sizing.scale400,
})}
>
<LabelSmall marginTop="0" marginBottom="0">
{isPendingProvision ? "Provisioning sandbox..." : "Creating session..."}
</LabelSmall>
<Skeleton rows={1} height="32px" />
<ParagraphSmall marginTop="0" marginBottom="0" color="contentSecondary">
{selectedForSession?.statusMessage ?? (isPendingProvision ? "The task is still provisioning." : "The session is being created.")}
</ParagraphSmall>
</div>
) : null}
{transcript.length === 0 && !(resolvedSessionId && sessionState.status === "loading") ? (
<EmptyState testId="session-transcript-empty">
{selectedForSession.runtimeStatus === "error" && selectedForSession.statusMessage
? `Session failed: ${selectedForSession.statusMessage}`
: !activeSandbox?.sandboxId
? selectedForSession.statusMessage
? `Sandbox unavailable: ${selectedForSession.statusMessage}`
: "This task is still provisioning its sandbox."
: staleSessionId
? `Session ${staleSessionId} is unavailable. Start a new session to continue.`
: resolvedSessionId
? "No transcript events yet. Send a prompt to start this session."
: "No active session for this task."}
: isPendingProvision
? (selectedForSession.statusMessage ?? "Provisioning sandbox...")
: isPendingSessionCreate
? "Creating session..."
: isSessionError
? (selectedSessionSummary?.errorMessage ?? "Session failed to start.")
: !activeSandbox?.sandboxId
? selectedForSession.statusMessage
? `Sandbox unavailable: ${selectedForSession.statusMessage}`
: "This task is still provisioning its sandbox."
: staleSessionId
? `Session ${staleSessionId} is unavailable. Start a new session to continue.`
: resolvedSessionId
? "No transcript events yet. Send a prompt to start this session."
: "No active session for this task."}
</EmptyState>
) : null}
@ -1442,7 +1475,7 @@ export function WorkspaceDashboard({ workspaceId, selectedTaskId, selectedRepoId
onChange={(event) => setDraft(event.target.value)}
placeholder="Send a follow-up prompt to this session"
rows={5}
disabled={!activeSandbox?.sandboxId}
disabled={!activeSandbox?.sandboxId || isPendingProvision || isPendingSessionCreate || isSessionError}
overrides={textareaTestIdOverrides("task-session-prompt")}
/>
<div
@ -1460,7 +1493,14 @@ export function WorkspaceDashboard({ workspaceId, selectedTaskId, selectedRepoId
void sendPrompt.mutateAsync(prompt);
}}
disabled={
sendPrompt.isPending || createSession.isPending || !selectedForSession || !activeSandbox?.sandboxId || draft.trim().length === 0
sendPrompt.isPending ||
createSession.isPending ||
!selectedForSession ||
!activeSandbox?.sandboxId ||
isPendingProvision ||
isPendingSessionCreate ||
isSessionError ||
draft.trim().length === 0
}
>
<span
@ -1837,7 +1877,7 @@ export function WorkspaceDashboard({ workspaceId, selectedTaskId, selectedRepoId
}}
data-testid="task-create-submit"
>
Create Task
{createTask.isPending ? "Creating..." : "Create Task"}
</Button>
</ModalFooter>
</Modal>

12
foundry/packages/shared/src/workbench.ts
View file

@ -1,8 +1,9 @@
import type { AgentType, ProviderId, TaskStatus } from "./contracts.js";
export type WorkbenchTaskStatus = "running" | "idle" | "new" | "archived";
export type WorkbenchTaskStatus = TaskStatus | "new";
export type WorkbenchAgentKind = "Claude" | "Codex" | "Cursor";
export type WorkbenchModelId = "claude-sonnet-4" | "claude-opus-4" | "gpt-4o" | "o3";
export type WorkbenchSessionStatus = "pending_provision" | "pending_session_create" | "ready" | "running" | "idle" | "error";
export interface WorkbenchTranscriptEvent {
id: string;
@ -27,10 +28,11 @@ export interface WorkbenchSessionSummary {
sessionName: string;
agent: WorkbenchAgentKind;
model: WorkbenchModelId;
status: "running" | "idle" | "error";
status: WorkbenchSessionStatus;
thinkingSinceMs: number | null;
unread: boolean;
created: boolean;
errorMessage?: string | null;
}
/** Full session content — only fetched when viewing a specific session tab. */
@ -42,10 +44,11 @@ export interface WorkbenchSessionDetail {
sessionName: string;
agent: WorkbenchAgentKind;
model: WorkbenchModelId;
status: "running" | "idle" | "error";
status: WorkbenchSessionStatus;
thinkingSinceMs: number | null;
unread: boolean;
created: boolean;
errorMessage?: string | null;
draft: WorkbenchComposerDraft;
transcript: WorkbenchTranscriptEvent[];
}
@ -166,6 +169,8 @@ export interface WorkbenchTask {
repoId: string;
title: string;
status: WorkbenchTaskStatus;
runtimeStatus?: TaskStatus;
statusMessage?: string | null;
repoName: string;
updatedAtMs: number;
branch: string | null;
@ -175,6 +180,7 @@ export interface WorkbenchTask {
diffs: Record<string, string>;
fileTree: WorkbenchFileTreeNode[];
minutesUsed: number;
activeSandboxId?: string | null;
}
export interface WorkbenchRepo {