* fix: pass resourceName to Azure provider from env or config
The azure custom loader was returning empty options, causing the
@ai-sdk/azure SDK to fail with 'Azure OpenAI resource name setting is
missing' when AZURE_RESOURCE_NAME is set in the environment.
Read resourceName from provider config options, AZURE_RESOURCE_NAME, or
AZURE_OPENAI_RESOURCE_NAME env vars (in that order of precedence) and
pass it explicitly to the SDK options, consistent with how
azure-cognitive-services handles its resource name.
Fixes#8273
* fix: also support baseURL for azure provider to avoid resourceName requirement
Users with a full endpoint URL (e.g. https://resource.openai.azure.com/openai/v1)
can now set baseURL in provider options or AZURE_OPENAI_ENDPOINT env var.
Per @ai-sdk/azure docs, baseURL takes precedence over resourceName and makes
it optional, fixing the 'resource name setting is missing' error for both
configuration styles.
* fix: add xhigh reasoning effort to Azure provider for GPT-5.4+
Azure variant logic was missing xhigh even though @ai-sdk/openai adds it
for models with release_date >= 2025-12-04. Azure GPT-5.4 now exposes
the same xhigh effort level as OpenAI/Kilo Gateway.
* fix(cli): annotate Azure reasoning changes
---------
Co-authored-by: marius-kilocode <marius@kilocode.ai>
* tweak: use theme tokens for debug bar surface
* chore: update nix node_modules hashes
* feat(tui): add heap snapshot functionality for TUI and server (#19028)
* ci
* change model for changelog
* release: v1.3.2
* fix(opencode): skip typechecking generated models snapshot (#19018)
* Revert "fix(app): more startup efficiency (#18985)"
This reverts commit 98b3340cee.
* Revert "fix(app): startup efficiency (#18854)"
This reverts commit 546748a461.
* effectify Worktree service (#18679)
* fix: increase operations-per-run to 1000 and pin stale action to v10.2.0
The stale-issues workflow was hitting the default 30 operations limit,
preventing it from processing all 2900+ issues/PRs. Increased to 1000
to handle the full backlog. Also pinned to exact v10.2.0 for reproducibility.
* Add close-issues script and GitHub Action
- Create script/github/close-issues.ts to close stale issues after 60 days
- Add GitHub Action workflow to run daily at 2 AM
- Remove old stale-issues workflow to avoid conflicts
* Fix close-issues workflow permissions
- Add contents: read permission for checkout
- Use github.token instead of secrets.GITHUB_TOKEN
* Process issues sequentially to avoid rate limits
* Change issue close reason from not_planned to completed
* fix(opencode): avoid snapshotting files over 2MB (#19043)
* fix: provide merge context to beta conflict resolver (#19055)
* tweak: only spawn lsp servers for files in current instance (or cwd if instance is global) (#19058)
* fix: beta resolver typecheck + build smoke check (#19060)
* fix: unblock beta conflict recovery (#19068)
* electron: add createDirectory to open directory picker (#19071)
* electron: remove file extension from electron-store wrapper (#19082)
* app: pre-warm project globalSync state when navigate project via keybind (#19088)
* fix(app): move message navigation off cmd+arrow (#18728)
* Reapply "fix(app): startup efficiency (#18854)"
This reverts commit a379eb3867.
* Reapply "fix(app): more startup efficiency (#18985)"
This reverts commit cbe1337f24.
* fix(app): hash inline script for csp
* Revert "fix(app): startup efficiency"
* Reapply "fix(app): startup efficiency"
This reverts commit 898456a25c.
* fix(app): opencode web server url
* chore(app): markdown playground in storyboard
* chore(app): markdown playground in storyboard
* feat(core): initial implementation of syncing (#17814)
* chore: generate
* chore: bump modelcontextprotocol/sdk to 1.27.1 (#19064)
* chore: storybook tweaks
* feat: restore git-backed review modes with effectful git service (#18900)
* chore: generate
* chore: update nix node_modules hashes
* chore: cleanup
* chore: remove dead code for todoread tool (#19128)
* chore: storybook tweaks
* fix(opencode): classify ZlibError from Bun fetch as retryable instead of unknown (#19104)
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com>
* fix(task): respect agent permission config for todowrite tool (#19125)
* fix(app): agent normalization (#19169)
* fix: Windows e2e stability (CrossSpawnSpawner, snapshot isolation, session race guards) (#19163)
* fix+refactor(mcp): lifecycle tests, cancelPending fix, Effect migration (#19042)
* effectify Bus service: migrate to Effect PubSub + InstanceState (#18579)
* file: use Effect.cached for scan deduplication (#19164)
* ignore: update disavowed list (#19184)
* skill: use Effect.cached for load deduplication (#19165)
* chore: generate
* fix: bump gitlab-ai-provider to 5.3.3 for DWS tool approval support (#19185)
* test: restore 5 workers on Windows e2e (#19188)
* fix(opencode): image paste on Windows Terminal 1.25+ with kitty keyboard (#17674)
* chore: update nix node_modules hashes
* wip: zen
* wip: zen
* go: do not respect disabled zen models
* fix: ensure enterprise url is set properly during auth flow (#19212)
* revert: roll back git-backed review modes (#19295)
* chore: generate
* tui: bypass local SSE event streaming in worker (#19183)
* feat: embed WebUI in binary with proxy flags (#19299)
Co-authored-by: BlankParticle <blankparticle@gmail.com>
* release: v1.3.3
* chore: generate
* changelog ci tweaks
* refactor(lsp): effectify LSP service with InstanceState (#19150)
* chore: generate
* feat: add gpt prompt so non codex gpt models have their own system prompt modeled after codex cli (#19220)
* feat(core): remove workspace server, WorkspaceContext, start work towards better routing (#19316)
* effectify Config service (#19139)
* chore: generate
* refactor(config): use cachedInvalidateWithTTL, bump effect to beta.37 (#19322)
* fix(mcp): close transport on failed/timed-out connections (#19200)
* fix(app): more startup perf (#19288)
* chore: generate
* chore: update nix node_modules hashes
* fix(app): don't bundle fonts (#19329)
* chore: generate
* fix(app): default shell tool to collapsed
* fix(app): remove fork session button
* fix(ui): reduce markdown jank while responses stream (#19304)
* fix: web ui bundle build on windows (#19337)
* refactor(effect): yield services instead of promise facades (#19325)
* chore: generate
* refactor(vcs): replace async git() with ChildProcessSpawner (#19361)
* fix(opencode): ignore generated models snapshot files (#19362)
* fix(ui): keep partial markdown readable while responses stream (#19403)
* chore: update nix node_modules hashes
* fix(app): persist queued followups across project switches (#19421)
* refactor(tool-registry): yield Config/Plugin services, use Effect.forEach (#19363)
* chore: generate
* tui plugins (#19347)
* chore: generate
* effectify Skill service internals (#19364)
* chore: update nix node_modules hashes
* effectify Plugin service internals (#19365)
* refactor(core): split out instance and route through workspaces (#19335)
* chore(app): more spacing controls
* fix(ui): make streamed markdown feel more continuous (#19404)
* fix(app): resize layout viewport when mobile keyboard appears (#15841)
* fix(desktop-electron): match dev dock icon inset on macOS (#19429)
* fix(app): default file tree to closed with minimum width (#19426)
* fix flaky plugin tests (no mock.module for bun) (#19445)
* tweak: add additional overflow error patterns (#19446)
* no theme override in dev (#19456)
* feat: AI SDK v6 support (#18433)
* refactor(session): effectify Session service (#19449)
* refactor(core): move more responsibility to workspace routing (#19455)
* chore: update nix node_modules hashes
* refactor(format): use ChildProcessSpawner instead of Process.spawn (#19457)
* chore: generate
* Single target plugin entrypoints (#19467)
* refactor(session): effectify SessionCompaction service (#19459)
* feat(ci): use Azure Artifact Signing for Windows releases (#15201)
* fix(app): more startup efficiency (#19454)
* update effect to 4.0.0-beta.42 (#19484)
* chore: update nix node_modules hashes
* tweak: adjust bash tool description to increase cache hit rates between projects (#19487)
* refactor(session): move context into prompt footer (#19486)
* refactor(prompt): remove variant cycle display from footer (#19489)
* feat: add model variant selection dialog (#19488)
* fix: restore subagent footer and fix style guide violations (#19491)
* tweak(session): add top spacing and remove obsolete docs prompt
* upgrade opentui to 0.1.91 (#19440)
* refactor(file): use AppFileSystem instead of raw Filesystem (#19458)
* chore: generate
* chore: update nix node_modules hashes
* kv theme before default fallback (#19523)
* feat: open dialog for model variant selection instead of cycling (#19534)
* refactor(session): effectify session processor (#19485)
* feat: dialog variant menu and subagent improvements (#19537)
* use theme color for prompt placeholder (#19535)
* fix: update opencode-gitlab-auth to 2.0.1 (#19552)
* chore: update nix node_modules hashes
* prompt slot (#19563)
* fix: respect semver build identifiers for nix (#11915)
* fix: nix embedded web-ui support (#19561)
* ignore: kill todo (#19566)
* chore: update nix node_modules hashes
* wip: zen
* wip: zen
* zen: ZDR policy
* ci: cancel stale nix-hashes runs (#19571)
* release: v1.3.4
* refactor: kilo compat for v1.3.4
* fix: migration types
* refactor: upgrade kilo-gateway to ai sdk v6
* refactor: improve upstream merge script
* fix: fix some tests
* style(kilo-vscode): adjust indentation and formatting in parts-util and PopupSelector
Normalize boolean expression indentation in isCompletionResult to use
consistent 4-space alignment and reformat PopupSelectorProps generic
interface declaration to split the Omit type across multiple lines.
* docs(kilo-docs): update auto-generated source links
Remove outdated URLs and add new bug report issue link pointing to
anomalyco/opencode repository. Drop references to kilocode bug report
template and config precedence order docs, reducing total unique URLs
from 262 to 261.
* fix(kilo-ui): remove unused NerdFonts story and MONO_NERD_FONTS import
Drop the NerdFonts story from font.stories.tsx along with the unused
MONO_NERD_FONTS import, as the exported constant is no longer available
from the @opencode-ai/ui/font module.
* chore: update visual regression baselines
* fix(opencode): move Show conditional wrapper outside box in home onboarding
Relocate the Show component to wrap the box element instead of being
nested inside it, preventing the empty box from rendering when the
onboarding tip is not visible.
---------
Co-authored-by: Jay V <air@live.ca>
Co-authored-by: opencode-agent[bot] <opencode-agent[bot]@users.noreply.github.com>
Co-authored-by: Dax <mail@thdxr.com>
Co-authored-by: Dax Raad <d@ironbay.co>
Co-authored-by: opencode <opencode@sst.dev>
Co-authored-by: Kit Langton <kit.langton@gmail.com>
Co-authored-by: Adam <2363879+adamdotdevin@users.noreply.github.com>
Co-authored-by: Luke Parker <10430890+Hona@users.noreply.github.com>
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com>
Co-authored-by: Brendan Allan <brendonovich@outlook.com>
Co-authored-by: Shoubhit Dash <shoubhit2005@gmail.com>
Co-authored-by: James Long <longster@gmail.com>
Co-authored-by: André Cruz <acruz@cloudflare.com>
Co-authored-by: Ariane Emory <97994360+ariane-emory@users.noreply.github.com>
Co-authored-by: Vladimir Glafirov <vglafirov@gitlab.com>
Co-authored-by: Frank <frank@anoma.ly>
Co-authored-by: BlankParticle <blankparticle@gmail.com>
Co-authored-by: Sebastian <hasta84@gmail.com>
Co-authored-by: Burak Yigit Kaya <byk@sentry.io>
Co-authored-by: Caleb Norton <n0603919@outlook.com>
Co-authored-by: Imanol Maiztegui <imanol.mzd@gmail.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Instant effort is not suitable for tool calling scenarios. Removed from
@ai-sdk/openai-compatible (direct inception) and updated kilo gateway
test to reflect server no longer providing instant variant. OpenRouter
keeps all OPENAI_EFFORTS unchanged.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add Mercury-2 reasoning effort variants for OpenRouter and Inception
direct API providers. OpenRouter gets OPENAI_EFFORTS (none through
xhigh), Inception gets instant + WIDELY_SUPPORTED_EFFORTS. Kilo
Gateway already works via server-provided variants.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Disable reasoning for Claude models via Kilo Gateway until the
OpenRouter AI SDK is upgraded to v2.x (which includes the fix for
reasoning_details duplication - OpenRouterTeam/ai-sdk-provider#344).
Changes:
- Remove hardcoded reasoning effort for Claude in options()
- Return empty variants for Claude in variants()
- Remove Claude-specific reasoning.effort in smallOptions()
- Keep thinking/reasoning block stripping code (prevents corrupt
reasoning_details from being sent back to the API)
- Caching continues to work via model.api.id matching
Fixes#303
* fix: use 'max' instead of 'xhigh' as highest reasoning effort for Anthropic models on Kilo Gateway
Anthropic's native effort levels are low/medium/high/max, not xhigh.
The upstream merge replaced the Kilo-specific effort mapping with
OPENAI_EFFORTS which uses xhigh. This restores 'max' as the
user-facing variant name while mapping it to 'xhigh' on the wire
for OpenRouter API compatibility.
* fix: clarify comment about OpenRouter effort naming vs Anthropic naming
* feat: enable reasoning for Claude models via Kilo Gateway
Enable extended thinking/reasoning support for Claude models when using
Kilo Gateway (OpenRouter-based). This includes:
- Add Kilo Gateway to OpenRouter SDK key mapping
- Skip toolCallId normalization for OpenRouter/Kilo to preserve thinking blocks
- Strip thinking/reasoning blocks from messages before sending to avoid
Anthropic API signature validation errors (blocks must be unchanged)
- Skip caching for OpenRouter/Kilo to avoid modifying thinking blocks
- Add Claude-specific reasoning effort variants for Kilo Gateway
- Enable default medium reasoning effort for Claude via Kilo Gateway
- Add proper smallOptions handling for Claude vs other models
The key insight is that Anthropic's API requires thinking blocks to be
EXACTLY unchanged with valid signatures. Since our storage reconstructs
messages, we strip thinking blocks (already shown to user) rather than
risk signature validation failures.
* fix: drop empty messages after stripping thinking blocks
Enable reasoning effort level cycling (Ctrl+T) for Claude models through
Kilo Gateway by:
- Adding @kilocode/kilo-gateway to sdkKey() to use 'openrouter' key for
provider options
- Adding dedicated variants() case for Claude/Anthropic models with full
effort levels (none, minimal, low, medium, high, xhigh)
- Supporting grok-3-mini reasoning variants through Kilo Gateway
- Fixing smallOptions() to use correct reasoning.effort format
OpenRouter maps effort levels to budget_tokens percentages:
- xhigh: 95%, high: 80%, medium: 50%, low: 20%, minimal: 10%, none: 0%
Imanol Maiztegui