Goose 1.41.0

[🔧 Improvements] MCP sampling support removed (#10087)

[🔧 Improvements] Legacy MCP-UI proxy support removed (#10086)

[🔧 Improvements] Session_id removed from provider streaming trait methods (#9984)

[🐛 Bug Fixes] Unused fetch-metadata IPC handler removed (SSRF) (#9340)

[✨ Features] `goose review` local code review command (#9114)

[✨ Features] Hooks feature (#9093)

[✨ Features] iFlytek Spark and Astron MaaS providers (#9837)

[✨ Features] `--edit` session flag to edit conversation before forking (#9799)

[✨ Features] Language selection feature in desktop (#9405)

[✨ Features] `/status` slash command in CLI (#9845)

[✨ Features] GLM-5.2 model support (#9967)

[✨ Features] Global hints loading from `~/.agents/AGENTS.md` (#9736)

[✨ Features] Azure Entra ID bearer token auth via `AZURE_OPENAI_AD_TOKEN` (#9716)

[✨ Features] MLX model support for local inference provider (#9154)

[✨ Features] Summon peek mode for async background tasks (#9519)

[✨ Features] Summon context parameter for delegate tasks (#9518)

[✨ Features] Summon working_dir override for delegate tasks (#9520)

[✨ Features] Exposed available tools in ACP schema (#10097)

[✨ Features] ACP thinking effort config option (#10711)

[✨ Features] Option to disable automatic update downloads (#9872)

[✨ Features] Cache token tracking for accurate cost reporting (#9752)

[✨ Features] Session import functionality (#9474)

[✨ Features] xAI SuperGrok OAuth subscription provider (#9420)

[✨ Features] Perplexity as declarative OpenAI-compatible provider (#9324)

[✨ Features] Alibaba (Qwen via DashScope) declarative provider (#9443)

[✨ Features] Configurable tool output size limit via `GOOSE_MAX_TOOL_RESPONSE_SIZE` (#9256)

[✨ Features] `/model` slash command for session model switching (#8747)

[✨ Features] Databricks AI Gateway provider (#9274)

[✨ Features] NEAR AI Cloud provider (#9352)

[✨ Features] Scaleway provider (#9254)

[✨ Features] `/goal` command for agent self-evaluation before finishing (#9069)

[✨ Features] Summon subagent instructions (#9325)

[✨ Features] Open-plugins generalization + skills (#9112)

[✨ Features] PreToolUse denial in hooks (#9304)

[✨ Features] Optional api_key configuration for declarative openai-engine providers (#9202)

[✨ Features] TUI diff viewer (#9260)

[✨ Features] Paginated session list in ACP (#9199)

[✨ Features] Structured per-provider config block with non-destructive provider switching (#8977)

[✨ Features] `tui` command on goose-cli (#9385)

[✨ Features] `goose://resume` session deep link (#9343)

[✨ Features] `GOOSE_FAST_MODEL` env var support (#9296)

[✨ Features] Vercel AI Gateway provider (#9144)

[✨ Features] Projects as backend sources with system prompt injection (#8739)

[✨ Features] Worktree-aware directory switcher (#8450)

[✨ Features] Chat history search feature (#8448)

[✨ Features] GitHub Copilot `/responses` API (#9043)

[✨ Features] Unified thinking effort control across all providers (#9242)

[✨ Features] Slash commands (built-in, skill, recipe) in ACP server (#9238)

[✨ Features] Atomic Chat as declarative OpenAI-compatible provider (#9210)

[✨ Features] Proactive OAuth token refresh to avoid re-auth on every session (#8386)

[✨ Features] Routstr as declarative provider (#9175)

[✨ Features] FuturMix provider (#8840)

[✨ Features] oMLX declarative provider (#9177)

[✨ Features] Image read tool (#9607)

[✨ Features] `gpt-5.5` model support for ChatGPT Codex provider (#9292)

[🔧 Improvements] Extended thinking disabled for fast-model operations (#9815)

[🔧 Improvements] MCP elicitations routed through tool streams (#9943)

[🔧 Improvements] MCP extensions supported in open plugins (#9471)

[🔧 Improvements] Hugging Face OAuth support added with auth tab in settings (#9552)

[✨ Features] Fireworks AI declarative provider (#9990)

[✨ Features] OpenRouter request parameters support (#9276)

[✨ Features] French, German, Italian, Portuguese, Indonesian, Malay, Vietnamese, and Traditional Chinese desktop locales (#10072)

[✨ Features] Spanish (es) locale support (#9833)

[✨ Features] Korean localization for desktop app (#9856)

[✨ Features] Japanese locale support (#9768)

[✨ Features] Hindi desktop locale (#9733)

[✨ Features] Streamable HTTP support in deeplink generator (#9954)

[✨ Features] Together AI declarative provider (#9545)

[✨ Features] OrcaRouter as declarative OpenAI-compatible provider (#9515)

[✨ Features] EmpirioLabs AI declarative provider (#9771)

[✨ Features] Structured metadata for summon task load results (#9521)

[✨ Features] TLS support for ACP serve (#10088)

[✨ Features] OpenAI Responses API store param configurability (#10040)

[✨ Features] Last message snippets for ACP sessions (#9798)

[✨ Features] OTLP logging schema for cross-tool detection (#9713)

[✨ Features] Russian language support (#9406)

[✨ Features] Turkish desktop locale (#9392)

[✨ Features] Harbor eval runner (#9138)

[✨ Features] Configurable `MAX_CODE_BLOCK_LINES` via env vars (#9301)

[✨ Features] Quarterly option for scheduler (#9076)

[✨ Features] Optional `--parameters` for scheduled recipes (#8741)

[✨ Features] Nushell terminal and completion support (#8628)

[✨ Features] `GOOSE_OAUTH_CALLBACK_PORT` for stable OAuth redirect_uri (#9209)

[✨ Features] `goose://new-session` deep link for fresh chat (#9196)

[✨ Features] ACP elicitation improvements (#9797)

[🔧 Improvements] Session name as source of truth for displayed session titles (#9841)

[🔧 Improvements] Databricks Responses API support check (#9347)

[🔧 Improvements] Initial prompt with `goose://new-session` (#9427)

[🔧 Improvements] View model interactions (#9806)

[🔧 Improvements] Generic validator for declarative providers (#10010)

[🔧 Improvements] LRU cache used for token counting (#9586)

[🔧 Improvements] Review subprocess timeout replaced with turn limits (#9571)

[🔧 Improvements] Canonical thinking modes added (#9743)

[🔧 Improvements] Skill instructions dynamically refreshed each turn (#9217)

[🔧 Improvements] Pasted HTML parsed better as markdown (#9190)

[🔧 Improvements] Summon instructions built per turn (#9329)

[🔧 Improvements] Settings moved into app shell (#9047)

[🔧 Improvements] Extension page redesigned (#8940)

[🔧 Improvements] Skills library redesigned (#8868)

[🔧 Improvements] Sidebar navigation and project icons polished (#8896)

[🔧 Improvements] Provider resolution parallelized and SQLite pool eagerly initialized (#8899)

[🔧 Improvements] Consecutive tool calls grouped into summarized chain card (#8995)

[🔧 Improvements] CLI help text updated for `--session-id` (#10077)

[🔧 Improvements] Canonical model registry refreshed (#10709, #9551, #9331)

[✨ Features] Simplified Chinese (zh-CN) translation (#8765)

[✨ Features] Declarative provider support in goose-providers crate (#9992)

[🔧 Improvements] EmpirioLabs provider display name updated (#10126)

[🐛 Bug Fixes] Handle resume deep links (#9298)

[🐛 Bug Fixes] Fall back to default provider when resuming session with unavailable provider (#9547)

[🐛 Bug Fixes] Bedrock real streaming via ConverseStream API (#9579)

[🐛 Bug Fixes] OAuth callback timeout in WSL instead of hanging (#9536)

[🐛 Bug Fixes] Login-shell PATH inherited in spawned subagents (#9737)

[🐛 Bug Fixes] Rejected OAuth credentials cleared after refresh (#9694)

[🐛 Bug Fixes] Anthropic stream refusals surfaced as visible errors (#9724)

[🐛 Bug Fixes] Deleting recently created session and creating new one no longer shows deleted content (#9351)

[🐛 Bug Fixes] Thinking config passed to Bedrock Anthropic models (#9794)

[🐛 Bug Fixes] Gateway respects `GOOSE_MAX_TURNS` (#9354)

[🐛 Bug Fixes] Per-session agent creation serialized to stop duplicate MCP init (#9357)

[🐛 Bug Fixes] OTLP signals skipped when protocol=grpc to avoid background-thread panic (#9512)

[🐛 Bug Fixes] Stop hook decisions honored when blocking (#9468)

[🐛 Bug Fixes] Telegram gateway error reporting and connection reliability improved (#9223)

[🐛 Bug Fixes] SSE reconnect loop kept alive on long disconnects (#8717)

[🐛 Bug Fixes] OpenAI request_params applied to outgoing API payload (#9151)

[🐛 Bug Fixes] Gemini 3.x known_location Global routing and KNOWN_MODELS entries (#9142)

[🐛 Bug Fixes] Anthropic Retry-After honored on 429 responses (#9161)

[🐛 Bug Fixes] Streaming Thinking deltas coalesced + available tools listed on not-found (#9162)

[🐛 Bug Fixes] OIDC-proxy exp enforcement independent of MAX_TOKEN_AGE_SECONDS (#8832)

[🐛 Bug Fixes] MCP-app host capabilities set correctly (#9116)

[🐛 Bug Fixes] Nullable schemas normalized for Vertex Gemini compatibility (#8930)

[🐛 Bug Fixes] 400 returned instead of panicking on invalid CSP header value (#8810)

[🐛 Bug Fixes] Tool name shown in approval prompt (#9216)

[🐛 Bug Fixes] Canonical model output paths (#10196)

[🐛 Bug Fixes] ACP agent cleanup errors propagation on session delete (#10112)

[🐛 Bug Fixes] `claude-sonnet-5` added to canonical model registry (#10183)

[🐛 Bug Fixes] MCP-app-guest request body limit increased (#10180)

[🐛 Bug Fixes] Tolerate partial Responses output items (#10108)

[🐛 Bug Fixes] Databricks tool parameters for OpenAI compatibility (#10101)

[🐛 Bug Fixes] Unescape shell-escaped image paths (#10098)

[🐛 Bug Fixes] Desktop Mac auto-update config seeding (#10132)

[🐛 Bug Fixes] `/edit` prompts saved to history (#10011)

[🐛 Bug Fixes] Schedule sessions listing using `session.message_count` (#10026)

[🐛 Bug Fixes] Session store sync after provider/model update and thinking effort update (#10060)

[🐛 Bug Fixes] Fallback REST API call removed in ACP mode for edit in place (#10034)

[🐛 Bug Fixes] Don't retry deterministically-permanent 400s (thinking-block immutability) (#10005)

[🐛 Bug Fixes] Async reqwest client for OTLP export in `goose serve` mode (#10100)

[🐛 Bug Fixes] Reject non-object tool-call arguments instead of panicking (#9832)

[🐛 Bug Fixes] Databricks pricing resolution and cost surfacing (#9925)

[🐛 Bug Fixes] Removed unused SubagentStart/SubagentStop hook events (#9960)

[🐛 Bug Fixes] Recipe modal closes when canceling parameter form (#9195)

[🐛 Bug Fixes] Tool request timestamp precedes tool response timestamp (#9462)

[🐛 Bug Fixes] Activate hermit on macOS so STDIO extensions resolve node (#9482)

[🐛 Bug Fixes] Auto-visualiser JSON-encoded string handling (#9838)

[🐛 Bug Fixes] User-Agent sent on URL image fetches (#9927)

[🐛 Bug Fixes] Do not pass temperature to ChatGPT Codex provider (#9931)

[🐛 Bug Fixes] Nostr session deep link import deduplication (#9918)

[🐛 Bug Fixes] Provider refresh when session working directory changes (#9883)

[🐛 Bug Fixes] GCP refreshed gcloud token after reauth (retry Vertex AI on 401/403) (#9849)

[🐛 Bug Fixes] Start a turn when `/goal` or `/grind` is set (#9801)

[🐛 Bug Fixes] PATH repair for plugin hook commands (#9615)

[🐛 Bug Fixes] Cmd/Ctrl+T new chat shortcut restored (#9614)

[🐛 Bug Fixes] Mention menu icon opens sidebar (#9643)

[🐛 Bug Fixes] Dedupe React to prevent multiple instances in renderer (#9631)

[🐛 Bug Fixes] Blocking OTLP HTTP exporter (#9599)

[🐛 Bug Fixes] Tokio runtime provided for OTLP HTTP exporter batch thread (#9541)

[🐛 Bug Fixes] OpenAI meta.n_ctx read from `/v1/models` for local server context limit (#9530)

[🐛 Bug Fixes] Session history scrollbar aligned with window edge (#9601)

[🐛 Bug Fixes] Session name refresh in sidebar after rename (#9543)

[🐛 Bug Fixes] Resolved skill supporting file paths shown (#9584)

[🐛 Bug Fixes] OpenAI-native cached_tokens recorded in usage metering (#9829)

[🐛 Bug Fixes] Moonshot provider API URL override option (#9819)

[🐛 Bug Fixes] Image paths with spaces detected (#9387)

[🐛 Bug Fixes] New chat shortcut navigation restored (#9528)

[🐛 Bug Fixes] Stale completed tasks evicted in summon (#9514)

[🐛 Bug Fixes] Unmatched extension names warning in delegate (#9513)

[🐛 Bug Fixes] Elicitation decline and cancel actions propagated end-to-end (#9437)

[🐛 Bug Fixes] Bedrock inference config (max_tokens, temperature) sent on Converse (#9889)

[🐛 Bug Fixes] Optimistic toggle state shown when disabling an extension (#9882)

[🐛 Bug Fixes] Windows backslash paths preserved in custom extension command (#9741)

[🐛 Bug Fixes] ACP cancel race condition fixed (#9804)

[🐛 Bug Fixes] Removed window goosed when closing window (#9818)

[🐛 Bug Fixes] Context exceeded checker made more precise (#9831)

[🐛 Bug Fixes] String values accepted for `GOOSE_CONTEXT_LIMIT` (#9738)

[🐛 Bug Fixes] Extended thinking kept within Anthropic output cap (#9814)

[🐛 Bug Fixes] Custom API path preserved in OpenAI base_url derivation (#9649)

[🐛 Bug Fixes] Bedrock ValidationException classified as ExecutionError (#9735)

[🐛 Bug Fixes] Unparseable extension entries preserved during config refresh (#9439)

[🐛 Bug Fixes] Session info ACP method (#9729)

[🐛 Bug Fixes] Ollama_cloud correctly mapped to canonical provider (#9639)

[🐛 Bug Fixes] Groq provider max tokens values corrected (#9790)

[🐛 Bug Fixes] Stream decode errors marked retryable (#9723)

[🐛 Bug Fixes] Databricks AI Gateway v2 endpoints pagination (#9753)

[🐛 Bug Fixes] Pattern-based prompt injection detection confidence scores adjusted (#9690)

[🐛 Bug Fixes] ACP compatibility in TUI for config extensions (#9683)

[🐛 Bug Fixes] Thinking content buffered across stream chunks for DeepSeek/Kimi tool calls (#9704)

[🐛 Bug Fixes] Actionable errors surfaced for truncated tool-call arguments (#9946)

[🐛 Bug Fixes] Custom OpenAI provider port preserved when URL scheme omitted (#9730)

[🐛 Bug Fixes] ACP agentInfo included in initialize response (#9765)

[🐛 Bug Fixes] Local inference media marker parts used for vision prompts (#9452)

[🐛 Bug Fixes] Extensions icon visible after switching sessions (#9787)

[🐛 Bug Fixes] Non-recipe project config files skipped in summon (#9808)

[🐛 Bug Fixes] Duplicate tool-call ids deduped within a turn (#9792)

[🐛 Bug Fixes] Bundled extensions resolved from discovery (#9759)

[🐛 Bug Fixes] LiteLLM context limit from `/model/info` for custom models (#9303)

[🐛 Bug Fixes] Thinking content preserved for provider context (#9314)

[🐛 Bug Fixes] Configurable timeout env vars mentioned in Ollama stream stall error (#9246)

[🐛 Bug Fixes] Full recipe parameter details included in load/discovery output (#9233)

[🐛 Bug Fixes] Anthropic empty object sent instead of null for tool_use input (#9355)

[🐛 Bug Fixes] Selected branch preserved across project chats (#9010)

[🐛 Bug Fixes] Tolerate missing responses output (#9449)

[🐛 Bug Fixes] Azure api-version query param made optional (#9221)

[🐛 Bug Fixes] Pending env vars preserved in Add Extension form (#9285)

[🐛 Bug Fixes] Linux x86_64 built inside manylinux_2_28 for glibc 2.28+ compat (#9415)

[🐛 Bug Fixes] Main window stops growing taller on every launch (#9409)

[🐛 Bug Fixes] Desktop chat search session limiting (#9366)

[🐛 Bug Fixes] ACP server context window size forwarded to clients (#9455)

[🐛 Bug Fixes] Provider list refresh in Switch Models picker (#9408)

[🐛 Bug Fixes] New chat started in current window from recipe param modal (#9422)

[🐛 Bug Fixes] Trace_output emitted as span attribute instead of event (#9255)

[🐛 Bug Fixes] File fallback checked when keyring has no entry (#9279)

[🐛 Bug Fixes] ScheduleModal error message styling (#9278)

[🐛 Bug Fixes] Sidebar hamburger aligned in macOS fullscreen (#9257)

[🐛 Bug Fixes] `current_exe()` used instead of PATH lookup when spawning goose (#9236)

[🐛 Bug Fixes] TCP_USER_TIMEOUT set on streamable HTTP clients (#9207)

[🐛 Bug Fixes] Custom provider activated after adding via configure (#9213)

[🐛 Bug Fixes] OTLP traces flushed reliably on exit with configurable timeout (#9228)

[🐛 Bug Fixes] Excessive MISSING_TRANSLATION warnings reduced for fallback locales (#9294)

[🐛 Bug Fixes] Cross-window deep link contamination eliminated (#9273)

[🐛 Bug Fixes] VT processing enabled on Windows Console Host (#9248)

[🐛 Bug Fixes] Cost zeroed for local providers (ollama, local) (#9222)

[🐛 Bug Fixes] Tool-use marker leakage prevented in toolshim output (#8310)

[🐛 Bug Fixes] Summon canonical limits re-applied when delegate overrides model (#9183)

[🐛 Bug Fixes] CLI cumulative total_tokens reported in stream-json/json output (#8910)

[🐛 Bug Fixes] GCP metadata server token refreshed on expiration (#8929)

[🐛 Bug Fixes] WSL2 OS detection corrected by removing PWD-based Windows override (#8869)

[🐛 Bug Fixes] Bedrock ReasoningContent blocks handled gracefully (#8843)

[🐛 Bug Fixes] Tool pair summaries inserted at chronological position in conversation (#9087)

[🐛 Bug Fixes] SQLite foreign key enforcement enabled in session pool (#9121)

[🐛 Bug Fixes] Deleted chat session still appearing in sidebar list (#8674)

[🐛 Bug Fixes] Extension_name required on read_resource (#8989)

[🐛 Bug Fixes] SACP notifies clients of generated session names (#8983)

[🐛 Bug Fixes] Quoted numeric config values converted to numbers if needed (#8844)

[🐛 Bug Fixes] User-set session name preserved for recipe-based chats (#9079)

[🐛 Bug Fixes] Inline code contrast improved in light theme (#9058)

[🐛 Bug Fixes] Max_tokens omitted for OpenAI-compatible requests when unset (#9123)

[🐛 Bug Fixes] GitHub recipe temp paths isolated (#8878)

[🐛 Bug Fixes] X-Initiator header set on GitHub Copilot requests (#8809)

[🐛 Bug Fixes] Dynamic_models: false honored in declarative provider configs (#8795)

[🐛 Bug Fixes] Frontend extension session state and tool routes (#8464)

[🐛 Bug Fixes] Bedrock trailing message cached for stable prefix across agent turns (#8916)

[🐛 Bug Fixes] Case-insensitive model name lookup for context_limit (#8906)

[🐛 Bug Fixes] Cumulative token usage emitted in stream-json/json complete event (#8870)

[🐛 Bug Fixes] ACP streaming chunks coalesced under one message id (#8788)

[🐛 Bug Fixes] Goose2 vite server reused on port conflict (#8722)

[🐛 Bug Fixes] Python3 used in developer extension instructions for macOS/Linux compatibility (#8784)

[🐛 Bug Fixes] ACP child synchronously reaped to avoid SIGCHLD race (#9023)

[🐛 Bug Fixes] Goose2 small-window chat and settings layouts (#9019)

[🐛 Bug Fixes] Elicitation fixes (#8999)

[🐛 Bug Fixes] OpenAI null tool_call arguments accepted in streaming chunks (#9035)

[🐛 Bug Fixes] Plain '> ' prompt used instead of goose emoji in CLI (#9305)

[🐛 Bug Fixes] Goosed not killed when a window closes (#9302)

[🐛 Bug Fixes] Databricks parallel tool image responses don't interleave tool results (#9241)

[🐛 Bug Fixes] Accumulated cost persisted in session DB to survive reload (#9191)

[🐛 Bug Fixes] Publish-npm binary built from current SHA + compat check (#9212)

[🐛 Bug Fixes] Preprompt excluded from session title generation (#8793)

[🐛 Bug Fixes] Login-shell PATH probe prevented from suspending goose on startup (#8804)

[🐛 Bug Fixes] Venice replaced with declarative config (#9234)

[🐛 Bug Fixes] MOIM allowlist widened for expected fix_conversation warnings (#9226)

[🐛 Bug Fixes] mkdir -p used for self-test workspace initialization (#9247)

[🐛 Bug Fixes] ACP images replayed on session load (#9496)

[🐛 Bug Fixes] Azure CLI resolved on Windows using az.cmd (#9215)

[🐛 Bug Fixes] Non-interactive terminal handled in goose configure on Windows (#9214)

[🐛 Bug Fixes] ACP recipe state applied during session load and fork (#9998)

[🐛 Bug Fixes] Local inference GGUF requirements stricter, auto detection of tool calling support, thinking output parsing fixed (#9442)

[🐛 Bug Fixes] Goose alias preferred for Databricks v2 inventory (#9430)

[🐛 Bug Fixes] Streamed tool calls without indexes supported (#10023)

[🐛 Bug Fixes] Allow local file ACP origins (#10194)

[🐛 Bug Fixes] MCP app sandbox bridge lifecycle fixed (#10064)

[🐛 Bug Fixes] Hermit cleanup added to uvx and jbang scripts (#9616)

[🐛 Bug Fixes] RPM build-id links disabled (#9671)

[🔧 Improvements] UI connected to ACP directly instead of goosed (#10081)

[🔧 Improvements] Refactored local inference provider crates (#10169)

[🔧 Improvements] ACP auth and origin defaults improved (#9886)

[🔧 Improvements] Turn-context kept out of Anthropic prefix cache (#10030)

[🔧 Improvements] Diagnostics migrated to JSON report (#9964)

[🔧 Improvements] Provider refactored: model config not required to create provider (#9953)

[🔧 Improvements] Provider inventory split out of providers (#9826)

[🔧 Improvements] API Client moved to goose-providers (#9869)

[🔧 Improvements] Request logger refactored out of providers (#9847)

[🔧 Improvements] Provider trait moved into goose-providers (#9860)

[🔧 Improvements] Global config access moved out of ModelConfig (#9769)

[🔧 Improvements] Debug info dropped for dependencies in dev profile (#9843)

[🔧 Improvements] Formats/openai.rs moved into goose-providers crate (#9633)

[🔧 Improvements] Goose-providers crate created with canonical models, conversation and other types (#9588)

[🔧 Improvements] UI types decoupled from generated OpenAPI types (#10149, #10109)

[🔧 Improvements] ACP SDK upgraded to new HTTP/WS crate (#10082)

[🔧 Improvements] Ollama provider moved into goose-providers (#9986)

[🔧 Improvements] Anthropic provider moved into goose-providers (#9985)

[🔧 Improvements] OpenAI provider moved into goose-providers (#9880)

[🔧 Improvements] Benchmarking improved (#9465, #9637)

[🔧 Improvements] Dependency hygiene improved (#9360)

[🔧 Improvements] Protocol cleanup (#9147)

[🔧 Improvements] Provider catalog moved behind ACP layer in goose2 (#9030)

[🔧 Improvements] Skill categories removed (#9008)

[🔧 Improvements] Artifact heuristics/regexes replaced with protocol messages (#8996)

[📚 Documentation] Custom agents guide (#9293)

[📚 Documentation] Custom agents guide correction (#9947)

[📚 Documentation] Message field named in hooks payload guide (#9913)

[📚 Documentation] Release testing instructions updated (#10122)

[📚 Documentation] ACP clients documentation updated (#9772)

[📚 Documentation] Session History navigation/delete docs fixed (#9727)

[📚 Documentation] Remote goosed server connection guide (#9275)

[📚 Documentation] SaladCloud AI Gateway documentation (#9253)

[📚 Documentation] Summon extension requirement documented for delegate and load tools (#9231)

[📚 Documentation] Hooks guide (#9288)

[📚 Documentation] Hooks feature blog post (#9227)

[📚 Documentation] Documentation reorganized (#9310)

[📚 Documentation] Scaleway provider documentation (#9423)

[📚 Documentation] Self-improving agents blog post (#9846)

[📚 Documentation] AAIF migration blog post (#9866)

[📚 Documentation] Orchestrating with goose blog post (#9104)

[📚 Documentation] Minimax and office QA benchmark findings blog post (#8984)

[📚 Documentation] Goose with peekaboo blog post (#8884)

[📚 Documentation] Built-in Local Inference blog post (#8808)

[📚 Documentation] DCO references removed (#9864)

[📚 Documentation] Internal documentation anchors fixed (#9094)

[📚 Documentation] Ophis MCP server added to extensions registry (#9861)

[📚 Documentation] Scholar Sidekick tutorial page (#9477)

[📚 Documentation] Installation whitespace fix (#9867)

[📚 Documentation] Stats update (#9410)

[📚 Documentation] MCP blog post typo fix (#9641)

[📚 Documentation] Analyze extension instructions updated (#9585)