refactor: finish companion rename migration

Complete the remaining pi-to-companion rename across companion-os, web, vm-orchestrator, docker, and archived fixtures.

Verification:
- semantic rg sweeps for Pi/piConfig/getPi/.pi runtime references
- npm run check in apps/companion-os (fails in this worktree: biome not found)

Co-authored-by: Codex <noreply@openai.com>

packages/ai/CHANGELOG.md

@@ -7,16 +7,16 @@
 ### Added
 
 - Added `gpt-5.4` model support for `openai`, `openai-codex`, `azure-openai-responses`, and `opencode` providers, with GPT-5.4 treated as xhigh-capable and capped to a 272000 context window in built-in metadata.
-- Added `gpt-5.3-codex` fallback model availability for `github-copilot` until upstream model catalogs include it ([#1853](https://github.com/badlogic/pi-mono/issues/1853)).
+- Added `gpt-5.3-codex` fallback model availability for `github-copilot` until upstream model catalogs include it ([#1853](https://github.com/badlogic/companion-mono/issues/1853)).
 
 ### Fixed
 
-- Preserved OpenAI Responses assistant `phase` metadata (`commentary`, `final_answer`) across turns by encoding `id` and `phase` in `textSignature` for session persistence and replay, with backward compatibility for legacy plain signatures ([#1819](https://github.com/badlogic/pi-mono/issues/1819)).
+- Preserved OpenAI Responses assistant `phase` metadata (`commentary`, `final_answer`) across turns by encoding `id` and `phase` in `textSignature` for session persistence and replay, with backward compatibility for legacy plain signatures ([#1819](https://github.com/badlogic/companion-mono/issues/1819)).
 - Fixed OpenAI Responses replay to omit empty thinking blocks, avoiding invalid no-op reasoning items in follow-up turns.
-- Switched the Mistral provider from the OpenAI-compatible completions path to Mistral's native SDK and conversations API, preserving native thinking blocks and Mistral-specific message semantics across turns ([#1716](https://github.com/badlogic/pi-mono/issues/1716)).
-- Fixed Antigravity endpoint fallback: 403/404 responses now cascade to the next endpoint instead of throwing immediately, added `autopush-cloudcode-pa.sandbox` endpoint to the fallback list, and removed extra fingerprint headers (`X-Goog-Api-Client`, `Client-Metadata`) from Antigravity requests ([#1830](https://github.com/badlogic/pi-mono/issues/1830)).
-- Fixed `@mariozechner/pi-ai/oauth` package exports to point directly at built `dist` files, avoiding broken TypeScript resolution through unpublished wrapper targets ([#1856](https://github.com/badlogic/pi-mono/issues/1856)).
-- Fixed Gemini 3 unsigned tool call replay: use `skip_thought_signature_validator` sentinel instead of converting function calls to text, preserving structured tool call context across multi-turn conversations ([#1829](https://github.com/badlogic/pi-mono/issues/1829)).
+- Switched the Mistral provider from the OpenAI-compatible completions path to Mistral's native SDK and conversations API, preserving native thinking blocks and Mistral-specific message semantics across turns ([#1716](https://github.com/badlogic/companion-mono/issues/1716)).
+- Fixed Antigravity endpoint fallback: 403/404 responses now cascade to the next endpoint instead of throwing immediately, added `autopush-cloudcode-pa.sandbox` endpoint to the fallback list, and removed extra fingerprint headers (`X-Goog-Api-Client`, `Client-Metadata`) from Antigravity requests ([#1830](https://github.com/badlogic/companion-mono/issues/1830)).
+- Fixed `@mariozechner/companion-ai/oauth` package exports to point directly at built `dist` files, avoiding broken TypeScript resolution through unpublished wrapper targets ([#1856](https://github.com/badlogic/companion-mono/issues/1856)).
+- Fixed Gemini 3 unsigned tool call replay: use `skip_thought_signature_validator` sentinel instead of converting function calls to text, preserving structured tool call context across multi-turn conversations ([#1829](https://github.com/badlogic/companion-mono/issues/1829)).
 
 ## [0.56.1] - 2026-03-05
 
@@ -24,12 +24,12 @@
 
 ### Breaking Changes
 
-- Moved Node OAuth runtime exports off the top-level package entry. Import OAuth login/refresh functions from `@mariozechner/pi-ai/oauth` instead of `@mariozechner/pi-ai` ([#1814](https://github.com/badlogic/pi-mono/issues/1814))
+- Moved Node OAuth runtime exports off the top-level package entry. Import OAuth login/refresh functions from `@mariozechner/companion-ai/oauth` instead of `@mariozechner/companion-ai` ([#1814](https://github.com/badlogic/companion-mono/issues/1814))
 
 ### Added
 
-- Added `gemini-3.1-flash-lite-preview` fallback model entry for the `google` provider so it remains selectable until upstream model catalogs include it ([#1785](https://github.com/badlogic/pi-mono/issues/1785), thanks [@n-WN](https://github.com/n-WN)).
-- Added OpenCode Go provider support with `opencode-go` model catalog entries and `OPENCODE_API_KEY` environment variable support ([#1757](https://github.com/badlogic/pi-mono/issues/1757)).
+- Added `gemini-3.1-flash-lite-preview` fallback model entry for the `google` provider so it remains selectable until upstream model catalogs include it ([#1785](https://github.com/badlogic/companion-mono/issues/1785), thanks [@n-WN](https://github.com/n-WN)).
+- Added OpenCode Go provider support with `opencode-go` model catalog entries and `OPENCODE_API_KEY` environment variable support ([#1757](https://github.com/badlogic/companion-mono/issues/1757)).
 
 ### Changed
 
@@ -37,11 +37,11 @@
 
 ### Fixed
 
-- Fixed Gemini 3.1 thinking-level detection in `google` and `google-vertex` providers so `gemini-3.1-*` models use Gemini 3 level-based thinking config instead of budget fallback ([#1785](https://github.com/badlogic/pi-mono/issues/1785), thanks [@n-WN](https://github.com/n-WN)).
-- Fixed browser bundling failures by lazy-loading the Bedrock provider and removing Node-only side effects from the default browser import graph ([#1814](https://github.com/badlogic/pi-mono/issues/1814)).
-- Fixed `ERR_VM_DYNAMIC_IMPORT_CALLBACK_MISSING` failures by replacing `Function`-based dynamic imports with module dynamic imports in browser-safe provider loading paths ([#1814](https://github.com/badlogic/pi-mono/issues/1814)).
-- Fixed Bedrock region resolution for `AWS_PROFILE` by honoring `region` from the selected profile when present ([#1800](https://github.com/badlogic/pi-mono/issues/1800)).
-- Fixed Groq Qwen3 reasoning effort mapping by translating unsupported effort values to provider-supported values ([#1745](https://github.com/badlogic/pi-mono/issues/1745)).
+- Fixed Gemini 3.1 thinking-level detection in `google` and `google-vertex` providers so `gemini-3.1-*` models use Gemini 3 level-based thinking config instead of budget fallback ([#1785](https://github.com/badlogic/companion-mono/issues/1785), thanks [@n-WN](https://github.com/n-WN)).
+- Fixed browser bundling failures by lazy-loading the Bedrock provider and removing Node-only side effects from the default browser import graph ([#1814](https://github.com/badlogic/companion-mono/issues/1814)).
+- Fixed `ERR_VM_DYNAMIC_IMPORT_CALLBACK_MISSING` failures by replacing `Function`-based dynamic imports with module dynamic imports in browser-safe provider loading paths ([#1814](https://github.com/badlogic/companion-mono/issues/1814)).
+- Fixed Bedrock region resolution for `AWS_PROFILE` by honoring `region` from the selected profile when present ([#1800](https://github.com/badlogic/companion-mono/issues/1800)).
+- Fixed Groq Qwen3 reasoning effort mapping by translating unsupported effort values to provider-supported values ([#1745](https://github.com/badlogic/companion-mono/issues/1745)).
 
 ## [0.55.4] - 2026-03-02
 
@@ -52,23 +52,23 @@
 ### Fixed
 
 - Restored built-in OAuth providers when unregistering dynamically registered provider IDs and added `resetOAuthProviders()` for registry reset flows.
-- Fixed Z.ai thinking control using wrong parameter name (`thinking` instead of `enable_thinking`), causing thinking to always be enabled and wasting tokens/latency ([#1674](https://github.com/badlogic/pi-mono/pull/1674) by [@okuyam2y](https://github.com/okuyam2y))
-- Fixed `redacted_thinking` blocks being silently dropped during Anthropic streaming. They are now captured as `ThinkingContent` with `redacted: true`, passed back to the API in multi-turn conversations, and handled in cross-model message transformation ([#1665](https://github.com/badlogic/pi-mono/pull/1665) by [@tctev](https://github.com/tctev))
-- Fixed `interleaved-thinking-2025-05-14` beta header being sent for adaptive thinking models (Opus 4.6, Sonnet 4.6) where the header is deprecated or redundant ([#1665](https://github.com/badlogic/pi-mono/pull/1665) by [@tctev](https://github.com/tctev))
-- Fixed temperature being sent alongside extended thinking, which is incompatible with both adaptive and budget-based thinking modes ([#1665](https://github.com/badlogic/pi-mono/pull/1665) by [@tctev](https://github.com/tctev))
-- Fixed `(external, cli)` user-agent flag causing 401 errors on Anthropic setup-token endpoint ([#1677](https://github.com/badlogic/pi-mono/pull/1677) by [@LazerLance777](https://github.com/LazerLance777))
-- Fixed crash when OpenAI-compatible provider returns a chunk with no `choices` array by adding optional chaining ([#1671](https://github.com/badlogic/pi-mono/issues/1671))
+- Fixed Z.ai thinking control using wrong parameter name (`thinking` instead of `enable_thinking`), causing thinking to always be enabled and wasting tokens/latency ([#1674](https://github.com/badlogic/companion-mono/pull/1674) by [@okuyam2y](https://github.com/okuyam2y))
+- Fixed `redacted_thinking` blocks being silently dropped during Anthropic streaming. They are now captured as `ThinkingContent` with `redacted: true`, passed back to the API in multi-turn conversations, and handled in cross-model message transformation ([#1665](https://github.com/badlogic/companion-mono/pull/1665) by [@tctev](https://github.com/tctev))
+- Fixed `interleaved-thinking-2025-05-14` beta header being sent for adaptive thinking models (Opus 4.6, Sonnet 4.6) where the header is deprecated or redundant ([#1665](https://github.com/badlogic/companion-mono/pull/1665) by [@tctev](https://github.com/tctev))
+- Fixed temperature being sent alongside extended thinking, which is incompatible with both adaptive and budget-based thinking modes ([#1665](https://github.com/badlogic/companion-mono/pull/1665) by [@tctev](https://github.com/tctev))
+- Fixed `(external, cli)` user-agent flag causing 401 errors on Anthropic setup-token endpoint ([#1677](https://github.com/badlogic/companion-mono/pull/1677) by [@LazerLance777](https://github.com/LazerLance777))
+- Fixed crash when OpenAI-compatible provider returns a chunk with no `choices` array by adding optional chaining ([#1671](https://github.com/badlogic/companion-mono/issues/1671))
 
 ## [0.55.1] - 2026-02-26
 
 ### Added
 
-- Added `gemini-3.1-pro-preview` model support to the `google-gemini-cli` provider ([#1599](https://github.com/badlogic/pi-mono/pull/1599) by [@audichuang](https://github.com/audichuang))
+- Added `gemini-3.1-pro-preview` model support to the `google-gemini-cli` provider ([#1599](https://github.com/badlogic/companion-mono/pull/1599) by [@audichuang](https://github.com/audichuang))
 
 ### Fixed
 
-- Fixed adaptive thinking for Claude Sonnet 4.6 in Anthropic and Bedrock providers, and clamped unsupported `xhigh` effort values to supported levels ([#1548](https://github.com/badlogic/pi-mono/pull/1548) by [@tctev](https://github.com/tctev))
-- Fixed Vertex ADC credential detection race by avoiding caching a false negative during async import initialization ([#1550](https://github.com/badlogic/pi-mono/pull/1550) by [@jeremiahgaylord-web](https://github.com/jeremiahgaylord-web))
+- Fixed adaptive thinking for Claude Sonnet 4.6 in Anthropic and Bedrock providers, and clamped unsupported `xhigh` effort values to supported levels ([#1548](https://github.com/badlogic/companion-mono/pull/1548) by [@tctev](https://github.com/tctev))
+- Fixed Vertex ADC credential detection race by avoiding caching a false negative during async import initialization ([#1550](https://github.com/badlogic/companion-mono/pull/1550) by [@jeremiahgaylord-web](https://github.com/jeremiahgaylord-web))
 
 ## [0.55.0] - 2026-02-24
 
@@ -108,16 +108,16 @@
 
 ### Added
 
-- Added optional `metadata` field to `StreamOptions` for passing provider-specific metadata (e.g. Anthropic `user_id` for abuse tracking/rate limiting) ([#1384](https://github.com/badlogic/pi-mono/pull/1384) by [@7Sageer](https://github.com/7Sageer))
+- Added optional `metadata` field to `StreamOptions` for passing provider-specific metadata (e.g. Anthropic `user_id` for abuse tracking/rate limiting) ([#1384](https://github.com/badlogic/companion-mono/pull/1384) by [@7Sageer](https://github.com/7Sageer))
 - Added `gpt-5.3-codex-spark` model definition for OpenAI and OpenAI Codex providers (128k context, text-only, research preview). Not yet functional, may become available in the next few hours or days.
 
 ### Changed
 
-- Routed GitHub Copilot Claude 4.x models through Anthropic Messages API, centralized Copilot dynamic header handling, and added Copilot Claude Anthropic stream coverage ([#1353](https://github.com/badlogic/pi-mono/pull/1353) by [@NateSmyth](https://github.com/NateSmyth))
+- Routed GitHub Copilot Claude 4.x models through Anthropic Messages API, centralized Copilot dynamic header handling, and added Copilot Claude Anthropic stream coverage ([#1353](https://github.com/badlogic/companion-mono/pull/1353) by [@NateSmyth](https://github.com/NateSmyth))
 
 ### Fixed
 
-- Fixed OpenAI completions and responses streams to tolerate malformed trailing tool-call JSON without failing parsing ([#1424](https://github.com/badlogic/pi-mono/issues/1424))
+- Fixed OpenAI completions and responses streams to tolerate malformed trailing tool-call JSON without failing parsing ([#1424](https://github.com/badlogic/companion-mono/issues/1424))
 
 ## [0.52.9] - 2026-02-08
 
@@ -127,31 +127,31 @@
 
 ### Fixed
 
-- Use `parametersJsonSchema` for Google provider tool declarations to support full JSON Schema (anyOf, oneOf, const, etc.) ([#1398](https://github.com/badlogic/pi-mono/issues/1398) by [@jarib](https://github.com/jarib))
+- Use `parametersJsonSchema` for Google provider tool declarations to support full JSON Schema (anyOf, oneOf, const, etc.) ([#1398](https://github.com/badlogic/companion-mono/issues/1398) by [@jarib](https://github.com/jarib))
 - Reverted incorrect Antigravity model change: `claude-opus-4-6-thinking` back to `claude-opus-4-5-thinking` (model doesn't exist on Antigravity endpoint)
-- Corrected opencode context windows for Claude Sonnet 4 and 4.5 ([#1383](https://github.com/badlogic/pi-mono/issues/1383))
+- Corrected opencode context windows for Claude Sonnet 4 and 4.5 ([#1383](https://github.com/badlogic/companion-mono/issues/1383))
 
 ## [0.52.8] - 2026-02-07
 
 ### Added
 
-- Added OpenRouter `auto` model alias for automatic model routing ([#1361](https://github.com/badlogic/pi-mono/pull/1361) by [@yogasanas](https://github.com/yogasanas))
+- Added OpenRouter `auto` model alias for automatic model routing ([#1361](https://github.com/badlogic/companion-mono/pull/1361) by [@yogasanas](https://github.com/yogasanas))
 
 ### Changed
 
-- Replaced Claude Opus 4.5 with Opus 4.6 in model definitions ([#1345](https://github.com/badlogic/pi-mono/pull/1345) by [@calvin-hpnet](https://github.com/calvin-hpnet))
+- Replaced Claude Opus 4.5 with Opus 4.6 in model definitions ([#1345](https://github.com/badlogic/companion-mono/pull/1345) by [@calvin-hpnet](https://github.com/calvin-hpnet))
 
 ## [0.52.7] - 2026-02-06
 
 ### Added
 
-- Added `AWS_BEDROCK_SKIP_AUTH` and `AWS_BEDROCK_FORCE_HTTP1` environment variables for connecting to unauthenticated Bedrock proxies ([#1320](https://github.com/badlogic/pi-mono/pull/1320) by [@virtuald](https://github.com/virtuald))
+- Added `AWS_BEDROCK_SKIP_AUTH` and `AWS_BEDROCK_FORCE_HTTP1` environment variables for connecting to unauthenticated Bedrock proxies ([#1320](https://github.com/badlogic/companion-mono/pull/1320) by [@virtuald](https://github.com/virtuald))
 
 ### Fixed
 
-- Set OpenAI Responses API requests to `store: false` by default to avoid server-side history logging ([#1308](https://github.com/badlogic/pi-mono/issues/1308))
-- Re-exported TypeBox `Type`, `Static`, and `TSchema` from `@mariozechner/pi-ai` to match documentation and avoid duplicate TypeBox type identity issues in pnpm setups ([#1338](https://github.com/badlogic/pi-mono/issues/1338))
-- Fixed Bedrock adaptive thinking handling for Claude Opus 4.6 with interleaved thinking beta responses ([#1323](https://github.com/badlogic/pi-mono/pull/1323) by [@markusylisiurunen](https://github.com/markusylisiurunen))
+- Set OpenAI Responses API requests to `store: false` by default to avoid server-side history logging ([#1308](https://github.com/badlogic/companion-mono/issues/1308))
+- Re-exported TypeBox `Type`, `Static`, and `TSchema` from `@mariozechner/companion-ai` to match documentation and avoid duplicate TypeBox type identity issues in pnpm setups ([#1338](https://github.com/badlogic/companion-mono/issues/1338))
+- Fixed Bedrock adaptive thinking handling for Claude Opus 4.6 with interleaved thinking beta responses ([#1323](https://github.com/badlogic/companion-mono/pull/1323) by [@markusylisiurunen](https://github.com/markusylisiurunen))
 - Fixed `AWS_BEDROCK_SKIP_AUTH` environment detection to avoid `process` access in non-Node.js environments
 
 ## [0.52.6] - 2026-02-05
@@ -201,13 +201,13 @@
 
 ### Fixed
 
-- Fixed OpenAI Codex Responses provider to respect configured baseUrl ([#1244](https://github.com/badlogic/pi-mono/issues/1244))
+- Fixed OpenAI Codex Responses provider to respect configured baseUrl ([#1244](https://github.com/badlogic/companion-mono/issues/1244))
 
 ## [0.51.5] - 2026-02-04
 
 ### Changed
 
-- Changed Bedrock model generation to drop legacy workarounds now handled upstream ([#1239](https://github.com/badlogic/pi-mono/pull/1239) by [@unexge](https://github.com/unexge))
+- Changed Bedrock model generation to drop legacy workarounds now handled upstream ([#1239](https://github.com/badlogic/companion-mono/pull/1239) by [@unexge](https://github.com/unexge))
 
 ## [0.51.4] - 2026-02-03
 
@@ -215,7 +215,7 @@
 
 ### Fixed
 
-- Fixed xhigh thinking level support check to accept gpt-5.2 model IDs ([#1209](https://github.com/badlogic/pi-mono/issues/1209))
+- Fixed xhigh thinking level support check to accept gpt-5.2 model IDs ([#1209](https://github.com/badlogic/companion-mono/issues/1209))
 
 ## [0.51.2] - 2026-02-03
 
@@ -229,23 +229,23 @@
 
 ### Fixed
 
-- Fixed `cacheRetention` option not being passed through in `buildBaseOptions` ([#1154](https://github.com/badlogic/pi-mono/issues/1154))
-- Fixed OAuth login/refresh not using HTTP proxy settings (`HTTP_PROXY`, `HTTPS_PROXY` env vars) ([#1132](https://github.com/badlogic/pi-mono/issues/1132))
-- Fixed OpenAI-compatible completions to omit unsupported `strict` tool fields for providers that reject them ([#1172](https://github.com/badlogic/pi-mono/issues/1172))
+- Fixed `cacheRetention` option not being passed through in `buildBaseOptions` ([#1154](https://github.com/badlogic/companion-mono/issues/1154))
+- Fixed OAuth login/refresh not using HTTP proxy settings (`HTTP_PROXY`, `HTTPS_PROXY` env vars) ([#1132](https://github.com/badlogic/companion-mono/issues/1132))
+- Fixed OpenAI-compatible completions to omit unsupported `strict` tool fields for providers that reject them ([#1172](https://github.com/badlogic/companion-mono/issues/1172))
 
 ## [0.50.9] - 2026-02-01
 
 ### Added
 
-- Added `PI_AI_ANTIGRAVITY_VERSION` environment variable to override the Antigravity User-Agent version when Google updates their version requirements ([#1129](https://github.com/badlogic/pi-mono/issues/1129))
-- Added `cacheRetention` stream option with provider-specific mappings for prompt cache controls, defaulting to short retention ([#1134](https://github.com/badlogic/pi-mono/issues/1134))
+- Added `COMPANION_AI_ANTIGRAVITY_VERSION` environment variable to override the Antigravity User-Agent version when Google updates their version requirements ([#1129](https://github.com/badlogic/companion-mono/issues/1129))
+- Added `cacheRetention` stream option with provider-specific mappings for prompt cache controls, defaulting to short retention ([#1134](https://github.com/badlogic/companion-mono/issues/1134))
 
 ## [0.50.8] - 2026-02-01
 
 ### Added
 
-- Added `maxRetryDelayMs` option to `StreamOptions` to cap server-requested retry delays. When a provider (e.g., Google Gemini CLI) requests a delay longer than this value, the request fails immediately with an informative error instead of waiting silently. Default: 60000ms (60 seconds). Set to 0 to disable the cap. ([#1123](https://github.com/badlogic/pi-mono/issues/1123))
-- Added Qwen thinking format support for OpenAI-compatible completions via `enable_thinking`. ([#940](https://github.com/badlogic/pi-mono/pull/940) by [@4h9fbZ](https://github.com/4h9fbZ))
+- Added `maxRetryDelayMs` option to `StreamOptions` to cap server-requested retry delays. When a provider (e.g., Google Gemini CLI) requests a delay longer than this value, the request fails immediately with an informative error instead of waiting silently. Default: 60000ms (60 seconds). Set to 0 to disable the cap. ([#1123](https://github.com/badlogic/companion-mono/issues/1123))
+- Added Qwen thinking format support for OpenAI-compatible completions via `enable_thinking`. ([#940](https://github.com/badlogic/companion-mono/pull/940) by [@4h9fbZ](https://github.com/4h9fbZ))
 
 ## [0.50.7] - 2026-01-31
 
@@ -257,12 +257,12 @@
 
 ### Added
 
-- Added Vercel AI Gateway routing support via `vercelGatewayRouting` option in model config ([#1051](https://github.com/badlogic/pi-mono/pull/1051) by [@ben-vargas](https://github.com/ben-vargas))
+- Added Vercel AI Gateway routing support via `vercelGatewayRouting` option in model config ([#1051](https://github.com/badlogic/companion-mono/pull/1051) by [@ben-vargas](https://github.com/ben-vargas))
 
 ### Fixed
 
-- Updated Antigravity User-Agent from 1.11.5 to 1.15.8 to fix rejected requests ([#1079](https://github.com/badlogic/pi-mono/issues/1079))
-- Fixed tool call argument defaults for Anthropic and Google history conversion when providers omit inputs ([#1065](https://github.com/badlogic/pi-mono/issues/1065))
+- Updated Antigravity User-Agent from 1.11.5 to 1.15.8 to fix rejected requests ([#1079](https://github.com/badlogic/companion-mono/issues/1079))
+- Fixed tool call argument defaults for Anthropic and Google history conversion when providers omit inputs ([#1065](https://github.com/badlogic/companion-mono/issues/1065))
 
 ## [0.50.3] - 2026-01-29
 
@@ -274,31 +274,31 @@
 
 ### Added
 
-- Added Hugging Face provider support via OpenAI-compatible Inference Router ([#994](https://github.com/badlogic/pi-mono/issues/994))
-- Added `PI_CACHE_RETENTION` environment variable to control cache TTL for Anthropic (5m vs 1h) and OpenAI (in-memory vs 24h). Set to `long` for extended retention. Only applies to direct API calls (api.anthropic.com, api.openai.com). ([#967](https://github.com/badlogic/pi-mono/issues/967))
+- Added Hugging Face provider support via OpenAI-compatible Inference Router ([#994](https://github.com/badlogic/companion-mono/issues/994))
+- Added `COMPANION_CACHE_RETENTION` environment variable to control cache TTL for Anthropic (5m vs 1h) and OpenAI (in-memory vs 24h). Set to `long` for extended retention. Only applies to direct API calls (api.anthropic.com, api.openai.com). ([#967](https://github.com/badlogic/companion-mono/issues/967))
 
 ### Fixed
 
-- Fixed OpenAI completions `toolChoice` handling to correctly set `type: "function"` wrapper ([#998](https://github.com/badlogic/pi-mono/pull/998) by [@williamtwomey](https://github.com/williamtwomey))
-- Fixed cross-provider handoff failing when switching from OpenAI Responses API providers (github-copilot, openai-codex) to other providers due to pipe-separated tool call IDs not being normalized, and trailing underscores in truncated IDs being rejected by OpenAI Codex ([#1022](https://github.com/badlogic/pi-mono/issues/1022))
-- Fixed 429 rate limit errors incorrectly triggering auto-compaction instead of retry with backoff ([#1038](https://github.com/badlogic/pi-mono/issues/1038))
-- Fixed Anthropic provider to handle `sensitive` stop_reason returned by API ([#978](https://github.com/badlogic/pi-mono/issues/978))
-- Fixed DeepSeek API compatibility by detecting `deepseek.com` URLs and disabling unsupported `developer` role ([#1048](https://github.com/badlogic/pi-mono/issues/1048))
-- Fixed Anthropic provider to preserve input token counts when proxies omit them in `message_delta` events ([#1045](https://github.com/badlogic/pi-mono/issues/1045))
+- Fixed OpenAI completions `toolChoice` handling to correctly set `type: "function"` wrapper ([#998](https://github.com/badlogic/companion-mono/pull/998) by [@williamtwomey](https://github.com/williamtwomey))
+- Fixed cross-provider handoff failing when switching from OpenAI Responses API providers (github-copilot, openai-codex) to other providers due to pipe-separated tool call IDs not being normalized, and trailing underscores in truncated IDs being rejected by OpenAI Codex ([#1022](https://github.com/badlogic/companion-mono/issues/1022))
+- Fixed 429 rate limit errors incorrectly triggering auto-compaction instead of retry with backoff ([#1038](https://github.com/badlogic/companion-mono/issues/1038))
+- Fixed Anthropic provider to handle `sensitive` stop_reason returned by API ([#978](https://github.com/badlogic/companion-mono/issues/978))
+- Fixed DeepSeek API compatibility by detecting `deepseek.com` URLs and disabling unsupported `developer` role ([#1048](https://github.com/badlogic/companion-mono/issues/1048))
+- Fixed Anthropic provider to preserve input token counts when proxies omit them in `message_delta` events ([#1045](https://github.com/badlogic/companion-mono/issues/1045))
 
 ## [0.50.1] - 2026-01-26
 
 ### Fixed
 
-- Fixed OpenCode Zen model generation to exclude deprecated models ([#970](https://github.com/badlogic/pi-mono/pull/970) by [@DanielTatarkin](https://github.com/DanielTatarkin))
+- Fixed OpenCode Zen model generation to exclude deprecated models ([#970](https://github.com/badlogic/companion-mono/pull/970) by [@DanielTatarkin](https://github.com/DanielTatarkin))
 
 ## [0.50.0] - 2026-01-26
 
 ### Added
 
-- Added OpenRouter provider routing support for custom models via `openRouterRouting` compat field ([#859](https://github.com/badlogic/pi-mono/pull/859) by [@v01dpr1mr0s3](https://github.com/v01dpr1mr0s3))
-- Added `azure-openai-responses` provider support for Azure OpenAI Responses API. ([#890](https://github.com/badlogic/pi-mono/pull/890) by [@markusylisiurunen](https://github.com/markusylisiurunen))
-- Added HTTP proxy environment variable support for API requests ([#942](https://github.com/badlogic/pi-mono/pull/942) by [@haoqixu](https://github.com/haoqixu))
+- Added OpenRouter provider routing support for custom models via `openRouterRouting` compat field ([#859](https://github.com/badlogic/companion-mono/pull/859) by [@v01dpr1mr0s3](https://github.com/v01dpr1mr0s3))
+- Added `azure-openai-responses` provider support for Azure OpenAI Responses API. ([#890](https://github.com/badlogic/companion-mono/pull/890) by [@markusylisiurunen](https://github.com/markusylisiurunen))
+- Added HTTP proxy environment variable support for API requests ([#942](https://github.com/badlogic/companion-mono/pull/942) by [@haoqixu](https://github.com/haoqixu))
 - Added `createAssistantMessageEventStream()` factory function for use in extensions.
 - Added `resetApiProviders()` to clear and re-register built-in API providers.
 
@@ -310,13 +310,13 @@
 
 ### Fixed
 
-- Fixed Bun runtime detection for dynamic imports in browser-compatible modules (stream.ts, openai-codex-responses.ts, openai-codex.ts) ([#922](https://github.com/badlogic/pi-mono/pull/922) by [@dannote](https://github.com/dannote))
+- Fixed Bun runtime detection for dynamic imports in browser-compatible modules (stream.ts, openai-codex-responses.ts, openai-codex.ts) ([#922](https://github.com/badlogic/companion-mono/pull/922) by [@dannote](https://github.com/dannote))
 - Fixed streaming functions to use `model.api` instead of hardcoded API types
 - Fixed Google providers to default tool call arguments to an empty object when omitted
-- Fixed OpenAI Responses streaming to handle `arguments.done` events on OpenAI-compatible endpoints ([#917](https://github.com/badlogic/pi-mono/pull/917) by [@williballenthin](https://github.com/williballenthin))
+- Fixed OpenAI Responses streaming to handle `arguments.done` events on OpenAI-compatible endpoints ([#917](https://github.com/badlogic/companion-mono/pull/917) by [@williballenthin](https://github.com/williballenthin))
 - Fixed OpenAI Codex Responses tool strictness handling after the shared responses refactor
 - Fixed Azure OpenAI Responses streaming to guard deltas before content parts and correct metadata and handoff gating
-- Fixed OpenAI completions tool-result image batching after consecutive tool results ([#902](https://github.com/badlogic/pi-mono/pull/902) by [@terrorobe](https://github.com/terrorobe))
+- Fixed OpenAI completions tool-result image batching after consecutive tool results ([#902](https://github.com/badlogic/companion-mono/pull/902) by [@terrorobe](https://github.com/terrorobe))
 
 ## [0.49.3] - 2026-01-22
 
@@ -324,21 +324,21 @@
 
 - Added `headers` option to `StreamOptions` for custom HTTP headers in API requests. Supported by all providers except Amazon Bedrock (which uses AWS SDK auth). Headers are merged with provider defaults and `model.headers`, with `options.headers` taking precedence.
 - Added `originator` option to `loginOpenAICodex()` for custom OAuth client identification
-- Browser compatibility for pi-ai: replaced top-level Node.js imports with dynamic imports for browser environments ([#873](https://github.com/badlogic/pi-mono/issues/873))
+- Browser compatibility for companion-ai: replaced top-level Node.js imports with dynamic imports for browser environments ([#873](https://github.com/badlogic/companion-mono/issues/873))
 
 ### Fixed
 
-- Fixed OpenAI Responses API 400 error "function_call without required reasoning item" when switching between models (same provider, different model). The fix omits the `id` field for function_calls from different models to avoid triggering OpenAI's reasoning/function_call pairing validation ([#886](https://github.com/badlogic/pi-mono/issues/886))
+- Fixed OpenAI Responses API 400 error "function_call without required reasoning item" when switching between models (same provider, different model). The fix omits the `id` field for function_calls from different models to avoid triggering OpenAI's reasoning/function_call pairing validation ([#886](https://github.com/badlogic/companion-mono/issues/886))
 
 ## [0.49.2] - 2026-01-19
 
 ### Added
 
-- Added AWS credential detection for ECS/Kubernetes environments: `AWS_CONTAINER_CREDENTIALS_RELATIVE_URI`, `AWS_CONTAINER_CREDENTIALS_FULL_URI`, `AWS_WEB_IDENTITY_TOKEN_FILE` ([#848](https://github.com/badlogic/pi-mono/issues/848))
+- Added AWS credential detection for ECS/Kubernetes environments: `AWS_CONTAINER_CREDENTIALS_RELATIVE_URI`, `AWS_CONTAINER_CREDENTIALS_FULL_URI`, `AWS_WEB_IDENTITY_TOKEN_FILE` ([#848](https://github.com/badlogic/companion-mono/issues/848))
 
 ### Fixed
 
-- Fixed OpenAI Responses 400 error "reasoning without following item" by skipping errored/aborted assistant messages entirely in transform-messages.ts ([#838](https://github.com/badlogic/pi-mono/pull/838))
+- Fixed OpenAI Responses 400 error "reasoning without following item" by skipping errored/aborted assistant messages entirely in transform-messages.ts ([#838](https://github.com/badlogic/companion-mono/pull/838))
 
 ### Removed
 
@@ -348,7 +348,7 @@
 
 ### Added
 
-- Added `OpenAIResponsesCompat` interface with `strictResponsesPairing` option for Azure OpenAI Responses API, which requires strict reasoning/message pairing in history replay ([#768](https://github.com/badlogic/pi-mono/pull/768) by [@prateekmedia](https://github.com/prateekmedia))
+- Added `OpenAIResponsesCompat` interface with `strictResponsesPairing` option for Azure OpenAI Responses API, which requires strict reasoning/message pairing in history replay ([#768](https://github.com/badlogic/companion-mono/pull/768) by [@prateekmedia](https://github.com/prateekmedia))
 
 ### Changed
 
@@ -356,7 +356,7 @@
 
 ### Fixed
 
-- Fixed tool call ID normalization for cross-provider handoffs (e.g., Codex to Antigravity Claude) ([#821](https://github.com/badlogic/pi-mono/issues/821))
+- Fixed tool call ID normalization for cross-provider handoffs (e.g., Codex to Antigravity Claude) ([#821](https://github.com/badlogic/companion-mono/issues/821))
 
 ## [0.49.0] - 2026-01-17
 
@@ -366,45 +366,45 @@
 
 ### Fixed
 
-- Fixed orphaned tool results after errored assistant messages causing Codex API errors. When an assistant message has `stopReason: "error"`, its tool calls are now excluded from pending tool tracking, preventing synthetic tool results from being generated for calls that will be dropped by provider-specific converters. ([#812](https://github.com/badlogic/pi-mono/issues/812))
-- Fixed Bedrock Claude max_tokens handling to always exceed thinking budget tokens, preventing compaction failures. ([#797](https://github.com/badlogic/pi-mono/pull/797) by [@pjtf93](https://github.com/pjtf93))
+- Fixed orphaned tool results after errored assistant messages causing Codex API errors. When an assistant message has `stopReason: "error"`, its tool calls are now excluded from pending tool tracking, preventing synthetic tool results from being generated for calls that will be dropped by provider-specific converters. ([#812](https://github.com/badlogic/companion-mono/issues/812))
+- Fixed Bedrock Claude max_tokens handling to always exceed thinking budget tokens, preventing compaction failures. ([#797](https://github.com/badlogic/companion-mono/pull/797) by [@pjtf93](https://github.com/pjtf93))
 - Fixed Claude Code tool name normalization to match the Claude Code tool list case-insensitively and remove invalid mappings.
 
 ## [0.48.0] - 2026-01-16
 
 ### Fixed
 
-- Fixed OpenAI-compatible provider feature detection to use `model.provider` in addition to URL, allowing custom base URLs (e.g., proxies) to work correctly with provider-specific settings ([#774](https://github.com/badlogic/pi-mono/issues/774))
+- Fixed OpenAI-compatible provider feature detection to use `model.provider` in addition to URL, allowing custom base URLs (e.g., proxies) to work correctly with provider-specific settings ([#774](https://github.com/badlogic/companion-mono/issues/774))
 - Fixed Gemini 3 context loss when switching from providers without thought signatures: unsigned tool calls are now converted to text with anti-mimicry notes instead of being skipped
-- Fixed string numbers in tool arguments not being coerced to numbers during validation ([#786](https://github.com/badlogic/pi-mono/pull/786) by [@dannote](https://github.com/dannote))
-- Fixed Bedrock tool call IDs to use only alphanumeric characters, avoiding API errors from invalid characters ([#781](https://github.com/badlogic/pi-mono/pull/781) by [@pjtf93](https://github.com/pjtf93))
+- Fixed string numbers in tool arguments not being coerced to numbers during validation ([#786](https://github.com/badlogic/companion-mono/pull/786) by [@dannote](https://github.com/dannote))
+- Fixed Bedrock tool call IDs to use only alphanumeric characters, avoiding API errors from invalid characters ([#781](https://github.com/badlogic/companion-mono/pull/781) by [@pjtf93](https://github.com/pjtf93))
 - Fixed empty error assistant messages (from 429/500 errors) breaking the tool_use to tool_result chain by filtering them in `transformMessages`
 
 ## [0.47.0] - 2026-01-16
 
 ### Fixed
 
-- Fixed OpenCode provider's `/v1` endpoint to use `system` role instead of `developer` role, fixing `400 Incorrect role information` error for models using `openai-completions` API ([#755](https://github.com/badlogic/pi-mono/pull/755) by [@melihmucuk](https://github.com/melihmucuk))
-- Added retry logic to OpenAI Codex provider for transient errors (429, 5xx, connection failures). Uses exponential backoff with up to 3 retries. ([#733](https://github.com/badlogic/pi-mono/issues/733))
+- Fixed OpenCode provider's `/v1` endpoint to use `system` role instead of `developer` role, fixing `400 Incorrect role information` error for models using `openai-completions` API ([#755](https://github.com/badlogic/companion-mono/pull/755) by [@melihmucuk](https://github.com/melihmucuk))
+- Added retry logic to OpenAI Codex provider for transient errors (429, 5xx, connection failures). Uses exponential backoff with up to 3 retries. ([#733](https://github.com/badlogic/companion-mono/issues/733))
 
 ## [0.46.0] - 2026-01-15
 
 ### Added
 
-- Added MiniMax China (`minimax-cn`) provider support ([#725](https://github.com/badlogic/pi-mono/pull/725) by [@tallshort](https://github.com/tallshort))
-- Added `gpt-5.2-codex` models for GitHub Copilot and OpenCode Zen providers ([#734](https://github.com/badlogic/pi-mono/pull/734) by [@aadishv](https://github.com/aadishv))
+- Added MiniMax China (`minimax-cn`) provider support ([#725](https://github.com/badlogic/companion-mono/pull/725) by [@tallshort](https://github.com/tallshort))
+- Added `gpt-5.2-codex` models for GitHub Copilot and OpenCode Zen providers ([#734](https://github.com/badlogic/companion-mono/pull/734) by [@aadishv](https://github.com/aadishv))
 
 ### Fixed
 
-- Avoid unsigned Gemini 3 tool calls ([#741](https://github.com/badlogic/pi-mono/pull/741) by [@roshanasingh4](https://github.com/roshanasingh4))
-- Fixed signature support for non-Anthropic models in Amazon Bedrock provider ([#727](https://github.com/badlogic/pi-mono/pull/727) by [@unexge](https://github.com/unexge))
+- Avoid unsigned Gemini 3 tool calls ([#741](https://github.com/badlogic/companion-mono/pull/741) by [@roshanasingh4](https://github.com/roshanasingh4))
+- Fixed signature support for non-Anthropic models in Amazon Bedrock provider ([#727](https://github.com/badlogic/companion-mono/pull/727) by [@unexge](https://github.com/unexge))
 
 ## [0.45.7] - 2026-01-13
 
 ### Fixed
 
-- Fixed OpenAI Responses timeout option handling ([#706](https://github.com/badlogic/pi-mono/pull/706) by [@markusylisiurunen](https://github.com/markusylisiurunen))
-- Fixed Bedrock tool call conversion to apply message transforms ([#707](https://github.com/badlogic/pi-mono/pull/707) by [@pjtf93](https://github.com/pjtf93))
+- Fixed OpenAI Responses timeout option handling ([#706](https://github.com/badlogic/companion-mono/pull/706) by [@markusylisiurunen](https://github.com/markusylisiurunen))
+- Fixed Bedrock tool call conversion to apply message transforms ([#707](https://github.com/badlogic/companion-mono/pull/707) by [@pjtf93](https://github.com/pjtf93))
 
 ## [0.45.6] - 2026-01-13
 
@@ -418,11 +418,11 @@
 
 ### Added
 
-- Added Vercel AI Gateway provider with model discovery and `AI_GATEWAY_API_KEY` env support ([#689](https://github.com/badlogic/pi-mono/pull/689) by [@timolins](https://github.com/timolins))
+- Added Vercel AI Gateway provider with model discovery and `AI_GATEWAY_API_KEY` env support ([#689](https://github.com/badlogic/companion-mono/pull/689) by [@timolins](https://github.com/timolins))
 
 ### Fixed
 
-- Fixed z.ai thinking/reasoning: z.ai uses `thinking: { type: "enabled" }` instead of OpenAI's `reasoning_effort`. Added `thinkingFormat` compat flag to handle this. ([#688](https://github.com/badlogic/pi-mono/issues/688))
+- Fixed z.ai thinking/reasoning: z.ai uses `thinking: { type: "enabled" }` instead of OpenAI's `reasoning_effort`. Added `thinkingFormat` compat flag to handle this. ([#688](https://github.com/badlogic/companion-mono/issues/688))
 
 ## [0.45.3] - 2026-01-13
 
@@ -434,11 +434,11 @@
 
 ### Added
 
-- MiniMax provider support with M2 and M2.1 models via Anthropic-compatible API ([#656](https://github.com/badlogic/pi-mono/pull/656) by [@dannote](https://github.com/dannote))
-- Add Amazon Bedrock provider with prompt caching for Claude models (experimental, tested with Anthropic Claude models only) ([#494](https://github.com/badlogic/pi-mono/pull/494) by [@unexge](https://github.com/unexge))
-- Added `serviceTier` option for OpenAI Responses requests ([#672](https://github.com/badlogic/pi-mono/pull/672) by [@markusylisiurunen](https://github.com/markusylisiurunen))
-- **Anthropic caching on OpenRouter**: Interactions with Anthropic models via OpenRouter now set a 5-minute cache point using Anthropic-style `cache_control` breakpoints on the last assistant or user message. ([#584](https://github.com/badlogic/pi-mono/pull/584) by [@nathyong](https://github.com/nathyong))
-- **Google Gemini CLI provider improvements**: Added Antigravity endpoint fallback (tries daily sandbox then prod when `baseUrl` is unset), header-based retry delay parsing (`Retry-After`, `x-ratelimit-reset`, `x-ratelimit-reset-after`), stable `sessionId` derivation from first user message for cache affinity, empty SSE stream retry with backoff, and `anthropic-beta` header for Claude thinking models ([#670](https://github.com/badlogic/pi-mono/pull/670) by [@kim0](https://github.com/kim0))
+- MiniMax provider support with M2 and M2.1 models via Anthropic-compatible API ([#656](https://github.com/badlogic/companion-mono/pull/656) by [@dannote](https://github.com/dannote))
+- Add Amazon Bedrock provider with prompt caching for Claude models (experimental, tested with Anthropic Claude models only) ([#494](https://github.com/badlogic/companion-mono/pull/494) by [@unexge](https://github.com/unexge))
+- Added `serviceTier` option for OpenAI Responses requests ([#672](https://github.com/badlogic/companion-mono/pull/672) by [@markusylisiurunen](https://github.com/markusylisiurunen))
+- **Anthropic caching on OpenRouter**: Interactions with Anthropic models via OpenRouter now set a 5-minute cache point using Anthropic-style `cache_control` breakpoints on the last assistant or user message. ([#584](https://github.com/badlogic/companion-mono/pull/584) by [@nathyong](https://github.com/nathyong))
+- **Google Gemini CLI provider improvements**: Added Antigravity endpoint fallback (tries daily sandbox then prod when `baseUrl` is unset), header-based retry delay parsing (`Retry-After`, `x-ratelimit-reset`, `x-ratelimit-reset-after`), stable `sessionId` derivation from first user message for cache affinity, empty SSE stream retry with backoff, and `anthropic-beta` header for Claude thinking models ([#670](https://github.com/badlogic/companion-mono/pull/670) by [@kim0](https://github.com/kim0))
 
 ## [0.44.0] - 2026-01-12
 
@@ -446,7 +446,7 @@
 
 ### Fixed
 
-- Fixed Google provider thinking detection: `isThinkingPart()` now only checks `thought === true`, not `thoughtSignature`. Per Google docs, `thoughtSignature` is for context replay and can appear on any part type. Also removed `id` field from `functionCall`/`functionResponse` (rejected by Vertex AI and Cloud Code Assist), and added `textSignature` round-trip for multi-turn reasoning context. ([#631](https://github.com/badlogic/pi-mono/pull/631) by [@theBucky](https://github.com/theBucky))
+- Fixed Google provider thinking detection: `isThinkingPart()` now only checks `thought === true`, not `thoughtSignature`. Per Google docs, `thoughtSignature` is for context replay and can appear on any part type. Also removed `id` field from `functionCall`/`functionResponse` (rejected by Vertex AI and Cloud Code Assist), and added `textSignature` round-trip for multi-turn reasoning context. ([#631](https://github.com/badlogic/companion-mono/pull/631) by [@theBucky](https://github.com/theBucky))
 
 ## [0.42.5] - 2026-01-11
 
@@ -456,21 +456,21 @@
 
 ### Changed
 
-- OpenAI Codex: switched to bundled system prompt matching opencode, changed originator to "pi", simplified prompt handling
+- OpenAI Codex: switched to bundled system prompt matching opencode, changed originator to "companion", simplified prompt handling
 
 ## [0.42.2] - 2026-01-10
 
 ### Added
 
 - Added `GOOGLE_APPLICATION_CREDENTIALS` env var support for Vertex AI credential detection (standard for CI/production).
-- Added `supportsUsageInStreaming` compatibility flag for OpenAI-compatible providers that reject `stream_options: { include_usage: true }`. Defaults to `true`. Set to `false` in model config for providers like gatewayz.ai. ([#596](https://github.com/badlogic/pi-mono/pull/596) by [@XesGaDeus](https://github.com/XesGaDeus))
-- Improved Google model pricing info ([#588](https://github.com/badlogic/pi-mono/pull/588) by [@aadishv](https://github.com/aadishv))
+- Added `supportsUsageInStreaming` compatibility flag for OpenAI-compatible providers that reject `stream_options: { include_usage: true }`. Defaults to `true`. Set to `false` in model config for providers like gatewayz.ai. ([#596](https://github.com/badlogic/companion-mono/pull/596) by [@XesGaDeus](https://github.com/XesGaDeus))
+- Improved Google model pricing info ([#588](https://github.com/badlogic/companion-mono/pull/588) by [@aadishv](https://github.com/aadishv))
 
 ### Fixed
 
 - Fixed `os.homedir()` calls at module load time; now resolved lazily when needed.
-- Fixed OpenAI Responses tool strict flag to use a boolean for LM Studio compatibility ([#598](https://github.com/badlogic/pi-mono/pull/598) by [@gnattu](https://github.com/gnattu))
-- Fixed Google Cloud Code Assist OAuth for paid subscriptions: properly handles long-running operations for project provisioning, supports `GOOGLE_CLOUD_PROJECT` / `GOOGLE_CLOUD_PROJECT_ID` env vars for paid tiers, and handles VPC-SC affected users ([#582](https://github.com/badlogic/pi-mono/pull/582) by [@cmf](https://github.com/cmf))
+- Fixed OpenAI Responses tool strict flag to use a boolean for LM Studio compatibility ([#598](https://github.com/badlogic/companion-mono/pull/598) by [@gnattu](https://github.com/gnattu))
+- Fixed Google Cloud Code Assist OAuth for paid subscriptions: properly handles long-running operations for project provisioning, supports `GOOGLE_CLOUD_PROJECT` / `GOOGLE_CLOUD_PROJECT_ID` env vars for paid tiers, and handles VPC-SC affected users ([#582](https://github.com/badlogic/companion-mono/pull/582) by [@cmf](https://github.com/cmf))
 
 ## [0.42.1] - 2026-01-09
 
@@ -492,24 +492,24 @@
 
 ### Fixed
 
-- Fixed Gemini CLI abort handling: detect native `AbortError` in retry catch block, cancel SSE reader when abort signal fires ([#568](https://github.com/badlogic/pi-mono/pull/568) by [@tmustier](https://github.com/tmustier))
-- Fixed Antigravity provider 429 errors by aligning request payload with CLIProxyAPI v6.6.89: inject Antigravity system instruction with `role: "user"`, set `requestType: "agent"`, and use `antigravity` userAgent. Added bridge prompt to override Antigravity behavior (identity, paths, web dev guidelines) with Pi defaults. ([#571](https://github.com/badlogic/pi-mono/pull/571) by [@ben-vargas](https://github.com/ben-vargas))
-- Fixed thinking block handling for cross-model conversations: thinking blocks are now converted to plain text (no `<thinking>` tags) when switching models. Previously, `<thinking>` tags caused models to mimic the pattern and output literal tags. Also fixed empty thinking blocks causing API errors. ([#561](https://github.com/badlogic/pi-mono/issues/561))
+- Fixed Gemini CLI abort handling: detect native `AbortError` in retry catch block, cancel SSE reader when abort signal fires ([#568](https://github.com/badlogic/companion-mono/pull/568) by [@tmustier](https://github.com/tmustier))
+- Fixed Antigravity provider 429 errors by aligning request payload with CLIProxyAPI v6.6.89: inject Antigravity system instruction with `role: "user"`, set `requestType: "agent"`, and use `antigravity` userAgent. Added bridge prompt to override Antigravity behavior (identity, paths, web dev guidelines) with companion defaults. ([#571](https://github.com/badlogic/companion-mono/pull/571) by [@ben-vargas](https://github.com/ben-vargas))
+- Fixed thinking block handling for cross-model conversations: thinking blocks are now converted to plain text (no `<thinking>` tags) when switching models. Previously, `<thinking>` tags caused models to mimic the pattern and output literal tags. Also fixed empty thinking blocks causing API errors. ([#561](https://github.com/badlogic/companion-mono/issues/561))
 
 ## [0.38.0] - 2026-01-08
 
 ### Added
 
-- `thinkingBudgets` option in `SimpleStreamOptions` for customizing token budgets per thinking level on token-based providers ([#529](https://github.com/badlogic/pi-mono/pull/529) by [@melihmucuk](https://github.com/melihmucuk))
+- `thinkingBudgets` option in `SimpleStreamOptions` for customizing token budgets per thinking level on token-based providers ([#529](https://github.com/badlogic/companion-mono/pull/529) by [@melihmucuk](https://github.com/melihmucuk))
 
 ### Breaking Changes
 
-- Removed OpenAI Codex model aliases (`gpt-5`, `gpt-5-mini`, `gpt-5-nano`, `codex-mini-latest`, `gpt-5-codex`, `gpt-5.1-codex`, `gpt-5.1-chat-latest`). Use canonical model IDs: `gpt-5.1`, `gpt-5.1-codex-max`, `gpt-5.1-codex-mini`, `gpt-5.2`, `gpt-5.2-codex`. ([#536](https://github.com/badlogic/pi-mono/pull/536) by [@ghoulr](https://github.com/ghoulr))
+- Removed OpenAI Codex model aliases (`gpt-5`, `gpt-5-mini`, `gpt-5-nano`, `codex-mini-latest`, `gpt-5-codex`, `gpt-5.1-codex`, `gpt-5.1-chat-latest`). Use canonical model IDs: `gpt-5.1`, `gpt-5.1-codex-max`, `gpt-5.1-codex-mini`, `gpt-5.2`, `gpt-5.2-codex`. ([#536](https://github.com/badlogic/companion-mono/pull/536) by [@ghoulr](https://github.com/ghoulr))
 
 ### Fixed
 
-- Fixed OpenAI Codex context window from 400,000 to 272,000 tokens to match Codex CLI defaults and prevent 400 errors. ([#536](https://github.com/badlogic/pi-mono/pull/536) by [@ghoulr](https://github.com/ghoulr))
-- Fixed Codex SSE error events to surface message, code, and status. ([#551](https://github.com/badlogic/pi-mono/pull/551) by [@tmustier](https://github.com/tmustier))
+- Fixed OpenAI Codex context window from 400,000 to 272,000 tokens to match Codex CLI defaults and prevent 400 errors. ([#536](https://github.com/badlogic/companion-mono/pull/536) by [@ghoulr](https://github.com/ghoulr))
+- Fixed Codex SSE error events to surface message, code, and status. ([#551](https://github.com/badlogic/companion-mono/pull/551) by [@tmustier](https://github.com/tmustier))
 - Fixed context overflow detection for `context_length_exceeded` error codes.
 
 ## [0.37.8] - 2026-01-07
@@ -520,7 +520,7 @@
 
 ### Added
 
-- Exported OpenAI Codex utilities: `CacheMetadata`, `getCodexInstructions`, `getModelFamily`, `ModelFamily`, `buildCodexPiBridge`, `buildCodexSystemPrompt`, `CodexSystemPrompt` ([#510](https://github.com/badlogic/pi-mono/pull/510) by [@mitsuhiko](https://github.com/mitsuhiko))
+- Exported OpenAI Codex utilities: `CacheMetadata`, `getCodexInstructions`, `getModelFamily`, `ModelFamily`, `buildCodexPiBridge`, `buildCodexSystemPrompt`, `CodexSystemPrompt` ([#510](https://github.com/badlogic/companion-mono/pull/510) by [@mitsuhiko](https://github.com/mitsuhiko))
 
 ## [0.37.5] - 2026-01-06
 
@@ -536,7 +536,7 @@
 
 ### Fixed
 
-- Codex provider now always includes `reasoning.encrypted_content` even when custom `include` options are passed ([#484](https://github.com/badlogic/pi-mono/pull/484) by [@kim0](https://github.com/kim0))
+- Codex provider now always includes `reasoning.encrypted_content` even when custom `include` options are passed ([#484](https://github.com/badlogic/companion-mono/pull/484) by [@kim0](https://github.com/kim0))
 
 ## [0.37.1] - 2026-01-05
 
@@ -544,22 +544,22 @@
 
 ### Breaking Changes
 
-- OpenAI Codex models no longer have per-thinking-level variants (e.g., `gpt-5.2-codex-high`). Use the base model ID and set thinking level separately. The Codex provider clamps reasoning effort to what each model supports internally. (initial implementation by [@ben-vargas](https://github.com/ben-vargas) in [#472](https://github.com/badlogic/pi-mono/pull/472))
+- OpenAI Codex models no longer have per-thinking-level variants (e.g., `gpt-5.2-codex-high`). Use the base model ID and set thinking level separately. The Codex provider clamps reasoning effort to what each model supports internally. (initial implementation by [@ben-vargas](https://github.com/ben-vargas) in [#472](https://github.com/badlogic/companion-mono/pull/472))
 
 ### Added
 
-- Headless OAuth support for all callback-server providers (Google Gemini CLI, Antigravity, OpenAI Codex): paste redirect URL when browser callback is unreachable ([#428](https://github.com/badlogic/pi-mono/pull/428) by [@ben-vargas](https://github.com/ben-vargas), [#468](https://github.com/badlogic/pi-mono/pull/468) by [@crcatala](https://github.com/crcatala))
+- Headless OAuth support for all callback-server providers (Google Gemini CLI, Antigravity, OpenAI Codex): paste redirect URL when browser callback is unreachable ([#428](https://github.com/badlogic/companion-mono/pull/428) by [@ben-vargas](https://github.com/ben-vargas), [#468](https://github.com/badlogic/companion-mono/pull/468) by [@crcatala](https://github.com/crcatala))
 - Cancellable GitHub Copilot device code polling via AbortSignal
 
 ### Fixed
 
-- Codex requests now omit the `reasoning` field entirely when thinking is off, letting the backend use its default instead of forcing a value. ([#472](https://github.com/badlogic/pi-mono/pull/472))
+- Codex requests now omit the `reasoning` field entirely when thinking is off, letting the backend use its default instead of forcing a value. ([#472](https://github.com/badlogic/companion-mono/pull/472))
 
 ## [0.36.0] - 2026-01-05
 
 ### Added
 
-- OpenAI Codex OAuth provider with Responses API streaming support: `openai-codex-responses` streaming provider with SSE parsing, tool-call handling, usage/cost tracking, and PKCE OAuth flow ([#451](https://github.com/badlogic/pi-mono/pull/451) by [@kim0](https://github.com/kim0))
+- OpenAI Codex OAuth provider with Responses API streaming support: `openai-codex-responses` streaming provider with SSE parsing, tool-call handling, usage/cost tracking, and PKCE OAuth flow ([#451](https://github.com/badlogic/companion-mono/pull/451) by [@kim0](https://github.com/kim0))
 
 ### Fixed
 
@@ -589,11 +589,11 @@
 
 ### Added
 
-- Vertex AI provider with ADC (Application Default Credentials) support. Authenticate with `gcloud auth application-default login`, set `GOOGLE_CLOUD_PROJECT` and `GOOGLE_CLOUD_LOCATION`, and access Gemini models via Vertex AI. ([#300](https://github.com/badlogic/pi-mono/pull/300) by [@default-anton](https://github.com/default-anton))
+- Vertex AI provider with ADC (Application Default Credentials) support. Authenticate with `gcloud auth application-default login`, set `GOOGLE_CLOUD_PROJECT` and `GOOGLE_CLOUD_LOCATION`, and access Gemini models via Vertex AI. ([#300](https://github.com/badlogic/companion-mono/pull/300) by [@default-anton](https://github.com/default-anton))
 
 ### Fixed
 
-- **Gemini CLI rate limit handling**: Added automatic retry with server-provided delay for 429 errors. Parses delay from error messages like "Your quota will reset after 39s" and waits accordingly. Falls back to exponential backoff for other transient errors. ([#370](https://github.com/badlogic/pi-mono/issues/370))
+- **Gemini CLI rate limit handling**: Added automatic retry with server-provided delay for 429 errors. Parses delay from error messages like "Your quota will reset after 39s" and waits accordingly. Falls back to exponential backoff for other transient errors. ([#370](https://github.com/badlogic/companion-mono/issues/370))
 
 ## [0.31.1] - 2026-01-02
 
@@ -601,7 +601,7 @@
 
 ### Breaking Changes
 
-- **Agent API moved**: All agent functionality (`agentLoop`, `agentLoopContinue`, `AgentContext`, `AgentEvent`, `AgentTool`, `AgentToolResult`, etc.) has moved to `@mariozechner/pi-agent-core`. Import from that package instead of `@mariozechner/pi-ai`.
+- **Agent API moved**: All agent functionality (`agentLoop`, `agentLoopContinue`, `AgentContext`, `AgentEvent`, `AgentTool`, `AgentToolResult`, etc.) has moved to `@mariozechner/companion-agent-core`. Import from that package instead of `@mariozechner/companion-ai`.
 
 ### Added
 
@@ -612,19 +612,19 @@
 ### Changed
 
 - **OAuth uses Web Crypto API**: PKCE generation and OAuth flows now use Web Crypto API (`crypto.subtle`) instead of Node.js `crypto` module. This improves browser compatibility while still working in Node.js 20+.
-- **Deterministic model generation**: `generate-models.ts` now sorts providers and models alphabetically for consistent output across runs. ([#332](https://github.com/badlogic/pi-mono/pull/332) by [@mrexodia](https://github.com/mrexodia))
+- **Deterministic model generation**: `generate-models.ts` now sorts providers and models alphabetically for consistent output across runs. ([#332](https://github.com/badlogic/companion-mono/pull/332) by [@mrexodia](https://github.com/mrexodia))
 
 ### Fixed
 
-- **OpenAI completions empty content blocks**: Empty text or thinking blocks in assistant messages are now filtered out before sending to the OpenAI completions API, preventing validation errors. ([#344](https://github.com/badlogic/pi-mono/pull/344) by [@default-anton](https://github.com/default-anton))
+- **OpenAI completions empty content blocks**: Empty text or thinking blocks in assistant messages are now filtered out before sending to the OpenAI completions API, preventing validation errors. ([#344](https://github.com/badlogic/companion-mono/pull/344) by [@default-anton](https://github.com/default-anton))
 - **Thinking token duplication**: Fixed thinking content duplication with chutes.ai provider. The provider was returning thinking content in both `reasoning_content` and `reasoning` fields, causing each chunk to be processed twice. Now only the first non-empty reasoning field is used.
-- **zAi provider API mapping**: Fixed zAi models to use `openai-completions` API with correct base URL (`https://api.z.ai/api/coding/paas/v4`) instead of incorrect Anthropic API mapping. ([#344](https://github.com/badlogic/pi-mono/pull/344), [#358](https://github.com/badlogic/pi-mono/pull/358) by [@default-anton](https://github.com/default-anton))
+- **zAi provider API mapping**: Fixed zAi models to use `openai-completions` API with correct base URL (`https://api.z.ai/api/coding/paas/v4`) instead of incorrect Anthropic API mapping. ([#344](https://github.com/badlogic/companion-mono/pull/344), [#358](https://github.com/badlogic/companion-mono/pull/358) by [@default-anton](https://github.com/default-anton))
 
 ## [0.28.0] - 2025-12-25
 
 ### Breaking Changes
 
-- **OAuth storage removed** ([#296](https://github.com/badlogic/pi-mono/issues/296)): All storage functions (`loadOAuthCredentials`, `saveOAuthCredentials`, `setOAuthStorage`, etc.) removed. Callers are responsible for storing credentials.
+- **OAuth storage removed** ([#296](https://github.com/badlogic/companion-mono/issues/296)): All storage functions (`loadOAuthCredentials`, `saveOAuthCredentials`, `setOAuthStorage`, etc.) removed. Callers are responsible for storing credentials.
 - **OAuth login functions**: `loginAnthropic`, `loginGitHubCopilot`, `loginGeminiCli`, `loginAntigravity` now return `OAuthCredentials` instead of saving to disk.
 - **refreshOAuthToken**: Now takes `(provider, credentials)` and returns new `OAuthCredentials` instead of saving.
 - **getOAuthApiKey**: Now takes `(provider, credentials)` and returns `{ newCredentials, apiKey }` or null.
@@ -636,13 +636,13 @@
 
 ### Fixed
 
-- **Thinking tag leakage**: Fixed Claude mimicking literal `</thinking>` tags in responses. Unsigned thinking blocks (from aborted streams) are now converted to plain text without `<thinking>` tags. The TUI still displays them as thinking blocks. ([#302](https://github.com/badlogic/pi-mono/pull/302) by [@nicobailon](https://github.com/nicobailon))
+- **Thinking tag leakage**: Fixed Claude mimicking literal `</thinking>` tags in responses. Unsigned thinking blocks (from aborted streams) are now converted to plain text without `<thinking>` tags. The TUI still displays them as thinking blocks. ([#302](https://github.com/badlogic/companion-mono/pull/302) by [@nicobailon](https://github.com/nicobailon))
 
 ## [0.25.1] - 2025-12-21
 
 ### Added
 
-- **xhigh thinking level support**: Added `supportsXhigh()` function to check if a model supports xhigh reasoning level. Also clamps xhigh to high for OpenAI models that don't support it. ([#236](https://github.com/badlogic/pi-mono/pull/236) by [@theBucky](https://github.com/theBucky))
+- **xhigh thinking level support**: Added `supportsXhigh()` function to check if a model supports xhigh reasoning level. Also clamps xhigh to high for OpenAI models that don't support it. ([#236](https://github.com/badlogic/companion-mono/pull/236) by [@theBucky](https://github.com/theBucky))
 
 ### Fixed
 
@@ -650,37 +650,37 @@
 
 - **Queued message steering**: When `getQueuedMessages` is provided, the agent loop now checks for queued user messages after each tool call and skips remaining tool calls in the current assistant message when a queued message arrives (emitting error tool results).
 
-- **Double API version path in Google provider URL**: Fixed Gemini API calls returning 404 after baseUrl support was added. The SDK was appending its default apiVersion to baseUrl which already included the version path. ([#251](https://github.com/badlogic/pi-mono/pull/251) by [@shellfyred](https://github.com/shellfyred))
+- **Double API version path in Google provider URL**: Fixed Gemini API calls returning 404 after baseUrl support was added. The SDK was appending its default apiVersion to baseUrl which already included the version path. ([#251](https://github.com/badlogic/companion-mono/pull/251) by [@shellfyred](https://github.com/shellfyred))
 
-- **Anthropic SDK retries disabled**: Re-enabled SDK-level retries (default 2) for transient HTTP failures. ([#252](https://github.com/badlogic/pi-mono/issues/252))
+- **Anthropic SDK retries disabled**: Re-enabled SDK-level retries (default 2) for transient HTTP failures. ([#252](https://github.com/badlogic/companion-mono/issues/252))
 
 ## [0.23.5] - 2025-12-19
 
 ### Added
 
-- **Gemini 3 Flash thinking support**: Extended thinking level support for Gemini 3 Flash models (MINIMAL, LOW, MEDIUM, HIGH) to match Pro models' capabilities. ([#212](https://github.com/badlogic/pi-mono/pull/212) by [@markusylisiurunen](https://github.com/markusylisiurunen))
+- **Gemini 3 Flash thinking support**: Extended thinking level support for Gemini 3 Flash models (MINIMAL, LOW, MEDIUM, HIGH) to match Pro models' capabilities. ([#212](https://github.com/badlogic/companion-mono/pull/212) by [@markusylisiurunen](https://github.com/markusylisiurunen))
 
-- **GitHub Copilot thinking models**: Added thinking support for additional Copilot models (o3-mini, o1-mini, o1-preview). ([#234](https://github.com/badlogic/pi-mono/pull/234) by [@aadishv](https://github.com/aadishv))
+- **GitHub Copilot thinking models**: Added thinking support for additional Copilot models (o3-mini, o1-mini, o1-preview). ([#234](https://github.com/badlogic/companion-mono/pull/234) by [@aadishv](https://github.com/aadishv))
 
 ### Fixed
 
-- **Gemini tool result format**: Fixed tool result format for Gemini 3 Flash Preview which strictly requires `{ output: value }` for success and `{ error: value }` for errors. Previous format using `{ result, isError }` was rejected by newer Gemini models. Also improved type safety by removing `as any` casts. ([#213](https://github.com/badlogic/pi-mono/issues/213), [#220](https://github.com/badlogic/pi-mono/pull/220))
+- **Gemini tool result format**: Fixed tool result format for Gemini 3 Flash Preview which strictly requires `{ output: value }` for success and `{ error: value }` for errors. Previous format using `{ result, isError }` was rejected by newer Gemini models. Also improved type safety by removing `as any` casts. ([#213](https://github.com/badlogic/companion-mono/issues/213), [#220](https://github.com/badlogic/companion-mono/pull/220))
 
-- **Google baseUrl configuration**: Google provider now respects `baseUrl` configuration for custom endpoints or API proxies. ([#216](https://github.com/badlogic/pi-mono/issues/216), [#221](https://github.com/badlogic/pi-mono/pull/221) by [@theBucky](https://github.com/theBucky))
+- **Google baseUrl configuration**: Google provider now respects `baseUrl` configuration for custom endpoints or API proxies. ([#216](https://github.com/badlogic/companion-mono/issues/216), [#221](https://github.com/badlogic/companion-mono/pull/221) by [@theBucky](https://github.com/theBucky))
 
-- **GitHub Copilot vision requests**: Added `Copilot-Vision-Request` header when sending images to GitHub Copilot models. ([#222](https://github.com/badlogic/pi-mono/issues/222))
+- **GitHub Copilot vision requests**: Added `Copilot-Vision-Request` header when sending images to GitHub Copilot models. ([#222](https://github.com/badlogic/companion-mono/issues/222))
 
-- **GitHub Copilot X-Initiator header**: Fixed X-Initiator logic to check last message role instead of any message in history. This ensures proper billing when users send follow-up messages. ([#209](https://github.com/badlogic/pi-mono/issues/209))
+- **GitHub Copilot X-Initiator header**: Fixed X-Initiator logic to check last message role instead of any message in history. This ensures proper billing when users send follow-up messages. ([#209](https://github.com/badlogic/companion-mono/issues/209))
 
 ## [0.22.3] - 2025-12-16
 
 ### Added
 
-- **Image limits test suite**: Added comprehensive tests for provider-specific image limitations (max images, max size, max dimensions). Discovered actual limits: Anthropic (100 images, 5MB, 8000px), OpenAI (500 images, ≥25MB), Gemini (~2500 images, ≥40MB), Mistral (8 images, ~15MB), OpenRouter (~40 images context-limited, ~15MB). ([#120](https://github.com/badlogic/pi-mono/pull/120))
+- **Image limits test suite**: Added comprehensive tests for provider-specific image limitations (max images, max size, max dimensions). Discovered actual limits: Anthropic (100 images, 5MB, 8000px), OpenAI (500 images, ≥25MB), Gemini (~2500 images, ≥40MB), Mistral (8 images, ~15MB), OpenRouter (~40 images context-limited, ~15MB). ([#120](https://github.com/badlogic/companion-mono/pull/120))
 
-- **Tool result streaming**: Added `tool_execution_update` event and optional `onUpdate` callback to `AgentTool.execute()` for streaming tool output during execution. Tools can now emit partial results (e.g., bash stdout) that are forwarded to subscribers. ([#44](https://github.com/badlogic/pi-mono/issues/44))
+- **Tool result streaming**: Added `tool_execution_update` event and optional `onUpdate` callback to `AgentTool.execute()` for streaming tool output during execution. Tools can now emit partial results (e.g., bash stdout) that are forwarded to subscribers. ([#44](https://github.com/badlogic/companion-mono/issues/44))
 
-- **X-Initiator header for GitHub Copilot**: Added X-Initiator header handling for GitHub Copilot provider to ensure correct call accounting (agent calls are not deducted from quota). Sets initiator based on last message role. ([#200](https://github.com/badlogic/pi-mono/pull/200) by [@kim0](https://github.com/kim0))
+- **X-Initiator header for GitHub Copilot**: Added X-Initiator header handling for GitHub Copilot provider to ensure correct call accounting (agent calls are not deducted from quota). Sets initiator based on last message role. ([#200](https://github.com/badlogic/companion-mono/pull/200) by [@kim0](https://github.com/kim0))
 
 ### Changed
 
@@ -688,7 +688,7 @@
 
 ### Fixed
 
-- **Reasoning disabled by default**: When `reasoning` option is not specified, thinking is now explicitly disabled for all providers. Previously, some providers like Gemini with "dynamic thinking" would use their default (thinking ON), causing unexpected token usage. This was the original intended behavior. ([#180](https://github.com/badlogic/pi-mono/pull/180) by [@markusylisiurunen](https://github.com/markusylisiurunen))
+- **Reasoning disabled by default**: When `reasoning` option is not specified, thinking is now explicitly disabled for all providers. Previously, some providers like Gemini with "dynamic thinking" would use their default (thinking ON), causing unexpected token usage. This was the original intended behavior. ([#180](https://github.com/badlogic/companion-mono/pull/180) by [@markusylisiurunen](https://github.com/markusylisiurunen))
 
 ## [0.22.2] - 2025-12-15
 
@@ -708,21 +708,21 @@ _Dedicated to Peter's shoulder ([@steipete](https://twitter.com/steipete))_
 
 ### Added
 
-- **GitHub Copilot provider**: Added `github-copilot` as a known provider with models sourced from models.dev. Includes Claude, GPT, Gemini, Grok, and other models available through GitHub Copilot. ([#191](https://github.com/badlogic/pi-mono/pull/191) by [@cau1k](https://github.com/cau1k))
+- **GitHub Copilot provider**: Added `github-copilot` as a known provider with models sourced from models.dev. Includes Claude, GPT, Gemini, Grok, and other models available through GitHub Copilot. ([#191](https://github.com/badlogic/companion-mono/pull/191) by [@cau1k](https://github.com/cau1k))
 
 ### Fixed
 
 - **GitHub Copilot gpt-5 models**: Fixed API selection for gpt-5 models to use `openai-responses` instead of `openai-completions` (gpt-5 models are not accessible via completions endpoint)
 
-- **GitHub Copilot cross-model context handoff**: Fixed context handoff failing when switching between GitHub Copilot models using different APIs (e.g., gpt-5 to claude-sonnet-4). Tool call IDs from OpenAI Responses API were incompatible with other models. ([#198](https://github.com/badlogic/pi-mono/issues/198))
+- **GitHub Copilot cross-model context handoff**: Fixed context handoff failing when switching between GitHub Copilot models using different APIs (e.g., gpt-5 to claude-sonnet-4). Tool call IDs from OpenAI Responses API were incompatible with other models. ([#198](https://github.com/badlogic/companion-mono/issues/198))
 
-- **Gemini 3 Pro thinking levels**: Thinking level configuration now works correctly for Gemini 3 Pro models. Previously all levels mapped to -1 (minimal thinking). Now LOW/MEDIUM/HIGH properly control test-time computation. ([#176](https://github.com/badlogic/pi-mono/pull/176) by [@markusylisiurunen](https://github.com/markusylisiurunen))
+- **Gemini 3 Pro thinking levels**: Thinking level configuration now works correctly for Gemini 3 Pro models. Previously all levels mapped to -1 (minimal thinking). Now LOW/MEDIUM/HIGH properly control test-time computation. ([#176](https://github.com/badlogic/companion-mono/pull/176) by [@markusylisiurunen](https://github.com/markusylisiurunen))
 
 ## [0.18.2] - 2025-12-11
 
 ### Changed
 
-- **Anthropic SDK retries disabled**: Set `maxRetries: 0` on Anthropic client to allow application-level retry handling. The SDK's built-in retries were interfering with coding-agent's retry logic. ([#157](https://github.com/badlogic/pi-mono/issues/157))
+- **Anthropic SDK retries disabled**: Set `maxRetries: 0` on Anthropic client to allow application-level retry handling. The SDK's built-in retries were interfering with coding-agent's retry logic. ([#157](https://github.com/badlogic/companion-mono/issues/157))
 
 ## [0.18.1] - 2025-12-10
 
@@ -732,11 +732,11 @@ _Dedicated to Peter's shoulder ([@steipete](https://twitter.com/steipete))_
 
 ### Fixed
 
-- Fixed Mistral 400 errors after aborted assistant messages by skipping empty assistant messages (no content, no tool calls) ([#165](https://github.com/badlogic/pi-mono/issues/165))
+- Fixed Mistral 400 errors after aborted assistant messages by skipping empty assistant messages (no content, no tool calls) ([#165](https://github.com/badlogic/companion-mono/issues/165))
 
-- Removed synthetic assistant bridge message after tool results for Mistral (no longer required as of Dec 2025) ([#165](https://github.com/badlogic/pi-mono/issues/165))
+- Removed synthetic assistant bridge message after tool results for Mistral (no longer required as of Dec 2025) ([#165](https://github.com/badlogic/companion-mono/issues/165))
 
-- Fixed bug where `ANTHROPIC_API_KEY` environment variable was deleted globally after first OAuth token usage, causing subsequent prompts to fail ([#164](https://github.com/badlogic/pi-mono/pull/164))
+- Fixed bug where `ANTHROPIC_API_KEY` environment variable was deleted globally after first OAuth token usage, causing subsequent prompts to fail ([#164](https://github.com/badlogic/companion-mono/pull/164))
 
 ## [0.17.0] - 2025-12-09
 
@@ -752,9 +752,9 @@ _Dedicated to Peter's shoulder ([@steipete](https://twitter.com/steipete))_
 
 - Added `validateToolCall(tools, toolCall)` helper that finds the tool by name and validates arguments.
 
-- **OpenAI compatibility overrides**: Added `compat` field to `Model` for `openai-completions` API, allowing explicit configuration of provider quirks (`supportsStore`, `supportsDeveloperRole`, `supportsReasoningEffort`, `maxTokensField`). Falls back to URL-based detection if not set. Useful for LiteLLM, custom proxies, and other non-standard endpoints. ([#133](https://github.com/badlogic/pi-mono/issues/133), thanks @fink-andreas for the initial idea and PR)
+- **OpenAI compatibility overrides**: Added `compat` field to `Model` for `openai-completions` API, allowing explicit configuration of provider quirks (`supportsStore`, `supportsDeveloperRole`, `supportsReasoningEffort`, `maxTokensField`). Falls back to URL-based detection if not set. Useful for LiteLLM, custom proxies, and other non-standard endpoints. ([#133](https://github.com/badlogic/companion-mono/issues/133), thanks @fink-andreas for the initial idea and PR)
 
-- **xhigh reasoning level**: Added `xhigh` to `ReasoningEffort` type for OpenAI codex-max models. For non-OpenAI providers (Anthropic, Google), `xhigh` is automatically mapped to `high`. ([#143](https://github.com/badlogic/pi-mono/issues/143))
+- **xhigh reasoning level**: Added `xhigh` to `ReasoningEffort` type for OpenAI codex-max models. For non-OpenAI providers (Anthropic, Google), `xhigh` is automatically mapped to `high`. ([#143](https://github.com/badlogic/companion-mono/issues/143))
 
 ### Changed
 

packages/ai/README.md

@@ -1,4 +1,4 @@
-# @mariozechner/pi-ai
+# @mariozechner/companion-ai
 
 Unified LLM API with automatic model discovery, provider configuration, token and cost tracking, and simple context persistence and hand-off to other models mid-session.
 
@@ -72,10 +72,10 @@ Unified LLM API with automatic model discovery, provider configuration, token an
 ## Installation
 
 ```bash
-npm install @mariozechner/pi-ai
+npm install @mariozechner/companion-ai
 ```
 
-TypeBox exports are re-exported from `@mariozechner/pi-ai`: `Type`, `Static`, and `TSchema`.
+TypeBox exports are re-exported from `@mariozechner/companion-ai`: `Type`, `Static`, and `TSchema`.
 
 ## Quick Start
 
@@ -88,7 +88,7 @@ import {
   Context,
   Tool,
   StringEnum,
-} from "@mariozechner/pi-ai";
+} from "@mariozechner/companion-ai";
 
 // Fully typed with auto-complete support for both providers and models
 const model = getModel("openai", "gpt-4o-mini");
@@ -223,7 +223,7 @@ Tools enable LLMs to interact with external systems. This library uses TypeBox s
 ### Defining Tools
 
 ```typescript
-import { Type, Tool, StringEnum } from "@mariozechner/pi-ai";
+import { Type, Tool, StringEnum } from "@mariozechner/companion-ai";
 
 // Define tool parameters with TypeBox
 const weatherTool: Tool = {
@@ -356,7 +356,7 @@ When using `agentLoop`, tool arguments are automatically validated against your
 When implementing your own tool execution loop with `stream()` or `complete()`, use `validateToolCall` to validate arguments before passing them to your tools:
 
 ```typescript
-import { stream, validateToolCall, Tool } from "@mariozechner/pi-ai";
+import { stream, validateToolCall, Tool } from "@mariozechner/companion-ai";
 
 const tools: Tool[] = [weatherTool, calculatorTool];
 const s = stream(model, { messages, tools });
@@ -410,7 +410,7 @@ Models with vision capabilities can process images. You can check if a model sup
 
 ```typescript
 import { readFileSync } from "fs";
-import { getModel, complete } from "@mariozechner/pi-ai";
+import { getModel, complete } from "@mariozechner/companion-ai";
 
 const model = getModel("openai", "gpt-4o-mini");
 
@@ -449,7 +449,7 @@ Many models support thinking/reasoning capabilities where they can show their in
 ### Unified Interface (streamSimple/completeSimple)
 
 ```typescript
-import { getModel, streamSimple, completeSimple } from "@mariozechner/pi-ai";
+import { getModel, streamSimple, completeSimple } from "@mariozechner/companion-ai";
 
 // Many models across providers support thinking/reasoning
 const model = getModel("anthropic", "claude-sonnet-4-20250514");
@@ -491,7 +491,7 @@ for (const block of response.content) {
 For fine-grained control, use the provider-specific options:
 
 ```typescript
-import { getModel, complete } from "@mariozechner/pi-ai";
+import { getModel, complete } from "@mariozechner/companion-ai";
 
 // OpenAI Reasoning (o1, o3, gpt-5)
 const openaiModel = getModel("openai", "gpt-5-mini");
@@ -578,7 +578,7 @@ if (message.stopReason === "error" || message.stopReason === "aborted") {
 The abort signal allows you to cancel in-progress requests. Aborted requests have `stopReason === 'aborted'`:
 
 ```typescript
-import { getModel, stream } from "@mariozechner/pi-ai";
+import { getModel, stream } from "@mariozechner/companion-ai";
 
 const model = getModel("openai", "gpt-4o-mini");
 const controller = new AbortController();
@@ -682,7 +682,7 @@ A **provider** offers models through a specific API. For example:
 ### Querying Providers and Models
 
 ```typescript
-import { getProviders, getModels, getModel } from "@mariozechner/pi-ai";
+import { getProviders, getModels, getModel } from "@mariozechner/companion-ai";
 
 // Get all available providers
 const providers = getProviders();
@@ -708,7 +708,7 @@ console.log(`Using ${model.name} via ${model.api} API`);
 You can create custom models for local inference servers or custom endpoints:
 
 ```typescript
-import { Model, stream } from "@mariozechner/pi-ai";
+import { Model, stream } from "@mariozechner/companion-ai";
 
 // Example: Ollama using OpenAI-compatible API
 const ollamaModel: Model<"openai-completions"> = {
@@ -802,7 +802,7 @@ If `compat` is not set, the library falls back to URL-based detection. If `compa
 Models are typed by their API, which keeps the model metadata accurate. Provider-specific option types are enforced when you call the provider functions directly. The generic `stream` and `complete` functions accept `StreamOptions` with additional provider fields.
 
 ```typescript
-import { streamAnthropic, type AnthropicOptions } from "@mariozechner/pi-ai";
+import { streamAnthropic, type AnthropicOptions } from "@mariozechner/companion-ai";
 
 // TypeScript knows this is an Anthropic model
 const claude = getModel("anthropic", "claude-sonnet-4-20250514");
@@ -831,7 +831,7 @@ When messages from one provider are sent to a different provider, the library au
 ### Example: Multi-Provider Conversation
 
 ```typescript
-import { getModel, complete, Context } from "@mariozechner/pi-ai";
+import { getModel, complete, Context } from "@mariozechner/companion-ai";
 
 // Start with Claude
 const claude = getModel("anthropic", "claude-sonnet-4-20250514");
@@ -884,7 +884,7 @@ This enables flexible workflows where you can:
 The `Context` object can be easily serialized and deserialized using standard JSON methods, making it simple to persist conversations, implement chat history, or transfer contexts between services:
 
 ```typescript
-import { Context, getModel, complete } from "@mariozechner/pi-ai";
+import { Context, getModel, complete } from "@mariozechner/companion-ai";
 
 // Create and use a context
 const context: Context = {
@@ -922,7 +922,7 @@ const continuation = await complete(newModel, restored);
 The library supports browser environments. You must pass the API key explicitly since environment variables are not available in browsers:
 
 ```typescript
-import { getModel, complete } from "@mariozechner/pi-ai";
+import { getModel, complete } from "@mariozechner/companion-ai";
 
 // API key must be passed explicitly in browser
 const model = getModel("anthropic", "claude-3-5-haiku-20241022");
@@ -943,7 +943,7 @@ const response = await complete(
 ### Browser Compatibility Notes
 
 - Amazon Bedrock (`bedrock-converse-stream`) is not supported in browser environments.
-- OAuth login flows are not supported in browser environments. Use the `@mariozechner/pi-ai/oauth` entry point in Node.js.
+- OAuth login flows are not supported in browser environments. Use the `@mariozechner/companion-ai/oauth` entry point in Node.js.
 - In browser builds, Bedrock can still appear in model lists. Calls to Bedrock models fail at runtime.
 - Use a server-side proxy or backend service if you need Bedrock or OAuth-based auth from a web app.
 
@@ -985,17 +985,17 @@ const response = await complete(model, context, {
 
 #### Antigravity Version Override
 
-Set `PI_AI_ANTIGRAVITY_VERSION` to override the Antigravity User-Agent version when Google updates their requirements:
+Set `COMPANION_AI_ANTIGRAVITY_VERSION` to override the Antigravity User-Agent version when Google updates their requirements:
 
 ```bash
-export PI_AI_ANTIGRAVITY_VERSION="1.23.0"
+export COMPANION_AI_ANTIGRAVITY_VERSION="1.23.0"
 ```
 
 #### Cache Retention
 
-Set `PI_CACHE_RETENTION=long` to extend prompt cache retention:
+Set `COMPANION_CACHE_RETENTION=long` to extend prompt cache retention:
 
-| Provider  | Default   | With `PI_CACHE_RETENTION=long` |
+| Provider  | Default   | With `COMPANION_CACHE_RETENTION=long` |
 | --------- | --------- | ------------------------------ |
 | Anthropic | 5 minutes | 1 hour                         |
 | OpenAI    | in-memory | 24 hours                       |
@@ -1007,7 +1007,7 @@ This only affects direct API calls to `api.anthropic.com` and `api.openai.com`.
 ### Checking Environment Variables
 
 ```typescript
-import { getEnvApiKey } from "@mariozechner/pi-ai";
+import { getEnvApiKey } from "@mariozechner/companion-ai";
 
 // Check if an API key is set in environment variables
 const key = getEnvApiKey("openai"); // checks OPENAI_API_KEY
@@ -1047,7 +1047,7 @@ export GOOGLE_APPLICATION_CREDENTIALS="/path/to/service-account.json"
 ```
 
 ```typescript
-import { getModel, complete } from "@mariozechner/pi-ai";
+import { getModel, complete } from "@mariozechner/companion-ai";
 
 (async () => {
   const model = getModel("google-vertex", "gemini-2.5-flash");
@@ -1068,16 +1068,16 @@ Official docs: [Application Default Credentials](https://cloud.google.com/docs/a
 The quickest way to authenticate:
 
 ```bash
-npx @mariozechner/pi-ai login              # interactive provider selection
-npx @mariozechner/pi-ai login anthropic    # login to specific provider
-npx @mariozechner/pi-ai list               # list available providers
+npx @mariozechner/companion-ai login              # interactive provider selection
+npx @mariozechner/companion-ai login anthropic    # login to specific provider
+npx @mariozechner/companion-ai list               # list available providers
 ```
 
 Credentials are saved to `auth.json` in the current directory.
 
 ### Programmatic OAuth
 
-The library provides login and token refresh functions via the `@mariozechner/pi-ai/oauth` entry point. Credential storage is the caller's responsibility.
+The library provides login and token refresh functions via the `@mariozechner/companion-ai/oauth` entry point. Credential storage is the caller's responsibility.
 
 ```typescript
 import {
@@ -1095,13 +1095,13 @@ import {
   // Types
   type OAuthProvider, // 'anthropic' | 'openai-codex' | 'github-copilot' | 'google-gemini-cli' | 'google-antigravity'
   type OAuthCredentials,
-} from "@mariozechner/pi-ai/oauth";
+} from "@mariozechner/companion-ai/oauth";
 ```
 
 ### Login Flow Example
 
 ```typescript
-import { loginGitHubCopilot } from "@mariozechner/pi-ai/oauth";
+import { loginGitHubCopilot } from "@mariozechner/companion-ai/oauth";
 import { writeFileSync } from "fs";
 
 const credentials = await loginGitHubCopilot({
@@ -1125,8 +1125,8 @@ writeFileSync("auth.json", JSON.stringify(auth, null, 2));
 Use `getOAuthApiKey()` to get an API key, automatically refreshing if expired:
 
 ```typescript
-import { getModel, complete } from "@mariozechner/pi-ai";
-import { getOAuthApiKey } from "@mariozechner/pi-ai/oauth";
+import { getModel, complete } from "@mariozechner/companion-ai";
+import { getOAuthApiKey } from "@mariozechner/companion-ai/oauth";
 import { readFileSync, writeFileSync } from "fs";
 
 // Load your stored credentials

packages/ai/package.json

@@ -1,5 +1,5 @@
 {
-  "name": "@mariozechner/pi-ai",
+  "name": "@mariozechner/companion-ai",
   "version": "0.56.2",
   "description": "Unified LLM API with automatic model discovery and provider configuration",
   "type": "module",
@@ -20,7 +20,7 @@
     }
   },
   "bin": {
-    "pi-ai": "./dist/cli.js"
+    "companion-ai": "./dist/cli.js"
   },
   "files": [
     "dist",

packages/ai/src/cli.ts

@@ -78,7 +78,7 @@ async function main(): Promise<void> {
     const providerList = PROVIDERS.map(
       (p) => `  ${p.id.padEnd(20)} ${p.name}`,
     ).join("\n");
-    console.log(`Usage: npx @mariozechner/pi-ai <command> [provider]
+    console.log(`Usage: npx @mariozechner/companion-ai <command> [provider]
 
 Commands:
   login [provider]  Login to an OAuth provider
@@ -88,9 +88,9 @@ Providers:
 ${providerList}
 
 Examples:
-  npx @mariozechner/pi-ai login              # interactive provider selection
-  npx @mariozechner/pi-ai login anthropic    # login to specific provider
-  npx @mariozechner/pi-ai list               # list providers
+  npx @mariozechner/companion-ai login              # interactive provider selection
+  npx @mariozechner/companion-ai login anthropic    # login to specific provider
+  npx @mariozechner/companion-ai list               # list providers
 `);
     return;
   }
@@ -131,7 +131,7 @@ Examples:
     if (!PROVIDERS.some((p) => p.id === provider)) {
       console.error(`Unknown provider: ${provider}`);
       console.error(
-        `Use 'npx @mariozechner/pi-ai list' to see available providers`,
+        `Use 'npx @mariozechner/companion-ai list' to see available providers`,
       );
       process.exit(1);
     }
@@ -142,7 +142,7 @@ Examples:
   }
 
   console.error(`Unknown command: ${command}`);
-  console.error(`Use 'npx @mariozechner/pi-ai --help' for usage`);
+  console.error(`Use 'npx @mariozechner/companion-ai --help' for usage`);
   process.exit(1);
 }
 

packages/ai/src/providers/amazon-bedrock.ts

@@ -514,7 +514,7 @@ function mapThinkingLevelToEffort(
 
 /**
  * Resolve cache retention preference.
- * Defaults to "short" and uses PI_CACHE_RETENTION for backward compatibility.
+ * Defaults to "short" and uses COMPANION_CACHE_RETENTION for backward compatibility.
  */
 function resolveCacheRetention(
   cacheRetention?: CacheRetention,
@@ -524,7 +524,7 @@ function resolveCacheRetention(
   }
   if (
     typeof process !== "undefined" &&
-    process.env.PI_CACHE_RETENTION === "long"
+    process.env.COMPANION_CACHE_RETENTION === "long"
   ) {
     return "long";
   }

packages/ai/src/providers/anthropic.ts

@@ -40,7 +40,7 @@ import { transformMessages } from "./transform-messages.js";
 
 /**
  * Resolve cache retention preference.
- * Defaults to "short" and uses PI_CACHE_RETENTION for backward compatibility.
+ * Defaults to "short" and uses COMPANION_CACHE_RETENTION for backward compatibility.
  */
 function resolveCacheRetention(
   cacheRetention?: CacheRetention,
@@ -50,7 +50,7 @@ function resolveCacheRetention(
   }
   if (
     typeof process !== "undefined" &&
-    process.env.PI_CACHE_RETENTION === "long"
+    process.env.COMPANION_CACHE_RETENTION === "long"
   ) {
     return "long";
   }

packages/ai/src/providers/google-gemini-cli.ts

@@ -88,7 +88,7 @@ const DEFAULT_ANTIGRAVITY_VERSION = "1.18.3";
 
 function getAntigravityHeaders() {
   const version =
-    process.env.PI_AI_ANTIGRAVITY_VERSION || DEFAULT_ANTIGRAVITY_VERSION;
+    process.env.COMPANION_AI_ANTIGRAVITY_VERSION || DEFAULT_ANTIGRAVITY_VERSION;
   return {
     "User-Agent": `antigravity/${version} darwin/arm64`,
   };
@@ -1040,8 +1040,8 @@ export function buildRequest(
     model: model.id,
     request,
     ...(isAntigravity ? { requestType: "agent" } : {}),
-    userAgent: isAntigravity ? "antigravity" : "pi-coding-agent",
-    requestId: `${isAntigravity ? "agent" : "pi"}-${Date.now()}-${Math.random().toString(36).slice(2, 11)}`,
+    userAgent: isAntigravity ? "antigravity" : "companion-coding-agent",
+    requestId: `${isAntigravity ? "agent" : "companion"}-${Date.now()}-${Math.random().toString(36).slice(2, 11)}`,
   };
 }
 

packages/ai/src/providers/openai-codex-responses.ts

@@ -997,10 +997,10 @@ function buildHeaders(
   headers.set("Authorization", `Bearer ${token}`);
   headers.set("chatgpt-account-id", accountId);
   headers.set("OpenAI-Beta", "responses=experimental");
-  headers.set("originator", "pi");
+  headers.set("originator", "companion");
   const userAgent = _os
-    ? `pi (${_os.platform()} ${_os.release()}; ${_os.arch()})`
-    : "pi (browser)";
+    ? `companion (${_os.platform()} ${_os.release()}; ${_os.arch()})`
+    : "companion (browser)";
   headers.set("User-Agent", userAgent);
   headers.set("accept", "text/event-stream");
   headers.set("content-type", "application/json");

packages/ai/src/providers/openai-responses.ts

@@ -33,7 +33,7 @@ const OPENAI_TOOL_CALL_PROVIDERS = new Set([
 
 /**
  * Resolve cache retention preference.
- * Defaults to "short" and uses PI_CACHE_RETENTION for backward compatibility.
+ * Defaults to "short" and uses COMPANION_CACHE_RETENTION for backward compatibility.
  */
 function resolveCacheRetention(
   cacheRetention?: CacheRetention,
@@ -43,7 +43,7 @@ function resolveCacheRetention(
   }
   if (
     typeof process !== "undefined" &&
-    process.env.PI_CACHE_RETENTION === "long"
+    process.env.COMPANION_CACHE_RETENTION === "long"
   ) {
     return "long";
   }

packages/ai/src/types.ts

@@ -283,7 +283,7 @@ export interface OpenAICompletionsCompat {
   supportsDeveloperRole?: boolean;
   /** Whether the provider supports `reasoning_effort`. Default: auto-detected from URL. */
   supportsReasoningEffort?: boolean;
-  /** Optional mapping from pi-ai reasoning levels to provider/model-specific `reasoning_effort` values. */
+  /** Optional mapping from companion-ai reasoning levels to provider/model-specific `reasoning_effort` values. */
   reasoningEffortMap?: Partial<Record<ThinkingLevel, string>>;
   /** Whether the provider supports `stream_options: { include_usage: true }` for token usage in streaming responses. Default: true. */
   supportsUsageInStreaming?: boolean;

packages/ai/src/utils/oauth/openai-codex.ts

@@ -216,7 +216,7 @@ async function refreshAccessToken(refreshToken: string): Promise<TokenResult> {
 }
 
 async function createAuthorizationFlow(
-  originator: string = "pi",
+  originator: string = "companion",
 ): Promise<{ verifier: string; state: string; url: string }> {
   const { verifier, challenge } = await generatePKCE();
   const state = createState();
@@ -337,7 +337,7 @@ function getAccountId(accessToken: string): string | null {
  * @param options.onManualCodeInput - Optional promise that resolves with user-pasted code.
  *                                    Races with browser callback - whichever completes first wins.
  *                                    Useful for showing paste input immediately alongside browser flow.
- * @param options.originator - OAuth originator parameter (defaults to "pi")
+ * @param options.originator - OAuth originator parameter (defaults to "companion")
  */
 export async function loginOpenAICodex(options: {
   onAuth: (info: { url: string; instructions?: string }) => void;

packages/ai/test/anthropic-tool-name-normalization.test.ts

@@ -71,8 +71,8 @@ describe.skipIf(!oauthToken)("Anthropic OAuth tool name normalization", () => {
     expect(toolCallName).toBe("todowrite");
   });
 
-  it("should handle pi's built-in tools (read, write, edit, bash)", async () => {
-    // Pi's tools use lowercase names, CC uses PascalCase
+  it("should handle companion's built-in tools (read, write, edit, bash)", async () => {
+    // Companion's tools use lowercase names, CC uses PascalCase
     const readTool: Tool = {
       name: "read",
       description: "Read a file",
@@ -116,7 +116,7 @@ describe.skipIf(!oauthToken)("Anthropic OAuth tool name normalization", () => {
   });
 
   it("should NOT map find to Glob - find is not a CC tool name", async () => {
-    // Pi has a "find" tool, CC has "Glob" - these are DIFFERENT tools
+    // Companion has a "find" tool, CC has "Glob" - these are DIFFERENT tools
     // The old code incorrectly mapped find -> Glob, which broke the round-trip
     // because there's no tool named "glob" in context.tools
     const findTool: Tool = {

packages/ai/test/cache-retention.test.ts

@@ -3,18 +3,18 @@ import { getModel } from "../src/models.js";
 import { stream } from "../src/stream.js";
 import type { Context } from "../src/types.js";
 
-describe("Cache Retention (PI_CACHE_RETENTION)", () => {
-  const originalEnv = process.env.PI_CACHE_RETENTION;
+describe("Cache Retention (COMPANION_CACHE_RETENTION)", () => {
+  const originalEnv = process.env.COMPANION_CACHE_RETENTION;
 
   beforeEach(() => {
-    delete process.env.PI_CACHE_RETENTION;
+    delete process.env.COMPANION_CACHE_RETENTION;
   });
 
   afterEach(() => {
     if (originalEnv !== undefined) {
-      process.env.PI_CACHE_RETENTION = originalEnv;
+      process.env.COMPANION_CACHE_RETENTION = originalEnv;
     } else {
-      delete process.env.PI_CACHE_RETENTION;
+      delete process.env.COMPANION_CACHE_RETENTION;
     }
   });
 
@@ -25,7 +25,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
 
   describe("Anthropic Provider", () => {
     it.skipIf(!process.env.ANTHROPIC_API_KEY)(
-      "should use default cache TTL (no ttl field) when PI_CACHE_RETENTION is not set",
+      "should use default cache TTL (no ttl field) when COMPANION_CACHE_RETENTION is not set",
       async () => {
         const model = getModel("anthropic", "claude-3-5-haiku-20241022");
         let capturedPayload: any = null;
@@ -51,9 +51,9 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
     );
 
     it.skipIf(!process.env.ANTHROPIC_API_KEY)(
-      "should use 1h cache TTL when PI_CACHE_RETENTION=long",
+      "should use 1h cache TTL when COMPANION_CACHE_RETENTION=long",
       async () => {
-        process.env.PI_CACHE_RETENTION = "long";
+        process.env.COMPANION_CACHE_RETENTION = "long";
         const model = getModel("anthropic", "claude-3-5-haiku-20241022");
         let capturedPayload: any = null;
 
@@ -79,7 +79,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
     );
 
     it("should not add ttl when baseUrl is not api.anthropic.com", async () => {
-      process.env.PI_CACHE_RETENTION = "long";
+      process.env.COMPANION_CACHE_RETENTION = "long";
 
       // Create a model with a different baseUrl (simulating a proxy)
       const baseModel = getModel("anthropic", "claude-3-5-haiku-20241022");
@@ -210,7 +210,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
 
   describe("OpenAI Responses Provider", () => {
     it.skipIf(!process.env.OPENAI_API_KEY)(
-      "should not set prompt_cache_retention when PI_CACHE_RETENTION is not set",
+      "should not set prompt_cache_retention when COMPANION_CACHE_RETENTION is not set",
       async () => {
         const model = getModel("openai", "gpt-4o-mini");
         let capturedPayload: any = null;
@@ -232,9 +232,9 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
     );
 
     it.skipIf(!process.env.OPENAI_API_KEY)(
-      "should set prompt_cache_retention to 24h when PI_CACHE_RETENTION=long",
+      "should set prompt_cache_retention to 24h when COMPANION_CACHE_RETENTION=long",
       async () => {
-        process.env.PI_CACHE_RETENTION = "long";
+        process.env.COMPANION_CACHE_RETENTION = "long";
         const model = getModel("openai", "gpt-4o-mini");
         let capturedPayload: any = null;
 
@@ -255,7 +255,7 @@ describe("Cache Retention (PI_CACHE_RETENTION)", () => {
     );
 
     it("should not set prompt_cache_retention when baseUrl is not api.openai.com", async () => {
-      process.env.PI_CACHE_RETENTION = "long";
+      process.env.COMPANION_CACHE_RETENTION = "long";
 
       // Create a model with a different baseUrl (simulating a proxy)
       const baseModel = getModel("openai", "gpt-4o-mini");

packages/ai/test/context-overflow.test.ts

@@ -683,7 +683,7 @@ describe("Context overflow error handling", () => {
 
   // Check if ollama is installed and local LLM tests are enabled
   let ollamaInstalled = false;
-  if (!process.env.PI_NO_LOCAL_LLM) {
+  if (!process.env.COMPANION_NO_LOCAL_LLM) {
     try {
       execSync("which ollama", { stdio: "ignore" });
       ollamaInstalled = true;
@@ -785,7 +785,7 @@ describe("Context overflow error handling", () => {
   // =============================================================================
 
   let lmStudioRunning = false;
-  if (!process.env.PI_NO_LOCAL_LLM) {
+  if (!process.env.COMPANION_NO_LOCAL_LLM) {
     try {
       execSync(
         "curl -s --max-time 1 http://localhost:1234/v1/models > /dev/null",

packages/ai/test/cross-provider-handoff.test.ts

@@ -227,7 +227,7 @@ function dumpFailurePayload(params: {
   payload?: unknown;
   messages: Message[];
 }): void {
-  const filename = `/tmp/pi-handoff-${params.label}-${Date.now()}.json`;
+  const filename = `/tmp/companion-handoff-${params.label}-${Date.now()}.json`;
   const body = {
     label: params.label,
     error: params.error,

packages/ai/test/empty.test.ts

@@ -765,7 +765,7 @@ describe("AI Providers Empty Message Tests", () => {
   );
 
   // =========================================================================
-  // OAuth-based providers (credentials from ~/.pi/agent/oauth.json)
+  // OAuth-based providers (credentials from ~/.companion/agent/oauth.json)
   // =========================================================================
 
   describe("Anthropic OAuth Provider Empty Messages", () => {

packages/ai/test/image-tool-result.test.ts

@@ -476,7 +476,7 @@ describe("Tool Results with Images", () => {
   );
 
   // =========================================================================
-  // OAuth-based providers (credentials from ~/.pi/agent/oauth.json)
+  // OAuth-based providers (credentials from ~/.companion/agent/oauth.json)
   // =========================================================================
 
   describe("Anthropic OAuth Provider (claude-sonnet-4-5)", () => {
@@ -584,7 +584,7 @@ describe("Tool Results with Images", () => {
       },
     );
 
-    /** These two don't work, the model simply won't call the tool, works in pi
+    /** These two don't work, the model simply won't call the tool, works in companion
 		it.skipIf(!antigravityToken)(
 			"claude-sonnet-4-5 - should handle tool result with only image",
 			{ retry: 3, timeout: 30000 },

packages/ai/test/oauth.ts

@@ -1,5 +1,5 @@
 /**
- * Test helper for resolving API keys from ~/.pi/agent/auth.json
+ * Test helper for resolving API keys from ~/.companion/agent/auth.json
  *
  * Supports both API key and OAuth credentials.
  * OAuth tokens are automatically refreshed if expired and saved back to auth.json.
@@ -20,7 +20,7 @@ import type {
   OAuthProvider,
 } from "../src/utils/oauth/types.js";
 
-const AUTH_PATH = join(homedir(), ".pi", "agent", "auth.json");
+const AUTH_PATH = join(homedir(), ".companion", "agent", "auth.json");
 
 type ApiKeyCredential = {
   type: "api_key";
@@ -57,7 +57,7 @@ function saveAuthStorage(storage: AuthStorage): void {
 }
 
 /**
- * Resolve API key for a provider from ~/.pi/agent/auth.json
+ * Resolve API key for a provider from ~/.companion/agent/auth.json
  *
  * For API key credentials, returns the key directly.
  * For OAuth credentials, returns the access token (refreshing if expired and saving back).

packages/ai/test/openai-codex-stream.test.ts

@@ -6,22 +6,22 @@ import { streamOpenAICodexResponses } from "../src/providers/openai-codex-respon
 import type { Context, Model } from "../src/types.js";
 
 const originalFetch = global.fetch;
-const originalAgentDir = process.env.PI_CODING_AGENT_DIR;
+const originalAgentDir = process.env.COMPANION_CODING_AGENT_DIR;
 
 afterEach(() => {
   global.fetch = originalFetch;
   if (originalAgentDir === undefined) {
-    delete process.env.PI_CODING_AGENT_DIR;
+    delete process.env.COMPANION_CODING_AGENT_DIR;
   } else {
-    process.env.PI_CODING_AGENT_DIR = originalAgentDir;
+    process.env.COMPANION_CODING_AGENT_DIR = originalAgentDir;
   }
   vi.restoreAllMocks();
 });
 
 describe("openai-codex streaming", () => {
   it("streams SSE responses into AssistantMessageEventStream", async () => {
-    const tempDir = mkdtempSync(join(tmpdir(), "pi-codex-stream-"));
-    process.env.PI_CODING_AGENT_DIR = tempDir;
+    const tempDir = mkdtempSync(join(tmpdir(), "companion-codex-stream-"));
+    process.env.COMPANION_CODING_AGENT_DIR = tempDir;
 
     const payload = Buffer.from(
       JSON.stringify({
@@ -95,7 +95,7 @@ describe("openai-codex streaming", () => {
         expect(headers?.get("Authorization")).toBe(`Bearer ${token}`);
         expect(headers?.get("chatgpt-account-id")).toBe("acc_test");
         expect(headers?.get("OpenAI-Beta")).toBe("responses=experimental");
-        expect(headers?.get("originator")).toBe("pi");
+        expect(headers?.get("originator")).toBe("companion");
         expect(headers?.get("accept")).toBe("text/event-stream");
         expect(headers?.has("x-api-key")).toBe(false);
         return new Response(stream, {
@@ -149,8 +149,8 @@ describe("openai-codex streaming", () => {
   });
 
   it("sets conversation_id/session_id headers and prompt_cache_key when sessionId is provided", async () => {
-    const tempDir = mkdtempSync(join(tmpdir(), "pi-codex-stream-"));
-    process.env.PI_CODING_AGENT_DIR = tempDir;
+    const tempDir = mkdtempSync(join(tmpdir(), "companion-codex-stream-"));
+    process.env.COMPANION_CODING_AGENT_DIR = tempDir;
 
     const payload = Buffer.from(
       JSON.stringify({
@@ -272,8 +272,8 @@ describe("openai-codex streaming", () => {
   it.each(["gpt-5.3-codex", "gpt-5.4"])(
     "clamps %s minimal reasoning effort to low",
     async (modelId) => {
-      const tempDir = mkdtempSync(join(tmpdir(), "pi-codex-stream-"));
-      process.env.PI_CODING_AGENT_DIR = tempDir;
+      const tempDir = mkdtempSync(join(tmpdir(), "companion-codex-stream-"));
+      process.env.COMPANION_CODING_AGENT_DIR = tempDir;
 
       const payload = Buffer.from(
         JSON.stringify({
@@ -393,8 +393,8 @@ describe("openai-codex streaming", () => {
   );
 
   it("does not set conversation_id/session_id headers when sessionId is not provided", async () => {
-    const tempDir = mkdtempSync(join(tmpdir(), "pi-codex-stream-"));
-    process.env.PI_CODING_AGENT_DIR = tempDir;
+    const tempDir = mkdtempSync(join(tmpdir(), "companion-codex-stream-"));
+    process.env.COMPANION_CODING_AGENT_DIR = tempDir;
 
     const payload = Buffer.from(
       JSON.stringify({

packages/ai/test/stream.test.ts

@@ -1048,7 +1048,7 @@ describe("Generate E2E Tests", () => {
   );
 
   // =========================================================================
-  // OAuth-based providers (credentials from ~/.pi/agent/oauth.json)
+  // OAuth-based providers (credentials from ~/.companion/agent/oauth.json)
   // Tokens are resolved at module level (see oauthTokens above)
   // =========================================================================
 
@@ -1800,7 +1800,7 @@ describe("Generate E2E Tests", () => {
 
   // Check if ollama is installed and local LLM tests are enabled
   let ollamaInstalled = false;
-  if (!process.env.PI_NO_LOCAL_LLM) {
+  if (!process.env.COMPANION_NO_LOCAL_LLM) {
     try {
       execSync("which ollama", { stdio: "ignore" });
       ollamaInstalled = true;

packages/ai/test/tokens.test.ts

@@ -294,7 +294,7 @@ describe("Token Statistics on Abort", () => {
   );
 
   // =========================================================================
-  // OAuth-based providers (credentials from ~/.pi/agent/oauth.json)
+  // OAuth-based providers (credentials from ~/.companion/agent/oauth.json)
   // =========================================================================
 
   describe("Anthropic OAuth Provider", () => {

packages/ai/test/tool-call-id-normalization.test.ts

@@ -7,7 +7,7 @@
  * OpenAI Responses API generates IDs in format: {call_id}|{id}
  * where {id} can be 400+ chars with special characters (+, /, =).
  *
- * Regression test for: https://github.com/badlogic/pi-mono/issues/1022
+ * Regression test for: https://github.com/badlogic/companion-mono/issues/1022
  */
 
 import { Type } from "@sinclair/typebox";

packages/ai/test/tool-call-without-result.test.ts

@@ -324,7 +324,7 @@ describe("Tool Call Without Result Tests", () => {
   });
 
   // =========================================================================
-  // OAuth-based providers (credentials from ~/.pi/agent/oauth.json)
+  // OAuth-based providers (credentials from ~/.companion/agent/oauth.json)
   // =========================================================================
 
   describe("Anthropic OAuth Provider", () => {

packages/ai/test/unicode-surrogate.test.ts

@@ -472,7 +472,7 @@ describe("AI Providers Unicode Surrogate Pair Tests", () => {
   );
 
   // =========================================================================
-  // OAuth-based providers (credentials from ~/.pi/agent/oauth.json)
+  // OAuth-based providers (credentials from ~/.companion/agent/oauth.json)
   // =========================================================================
 
   describe("Anthropic OAuth Provider Unicode Handling", () => {