d97fb9a6da
fix: revert Claude reasoning additions for Kilo Gateway ( #305 )
...
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
2026-02-13 10:38:01 +01:00
52588386a4
feat: merge opencode v1.1.57
2026-02-11 14:26:25 -03:00
0fd6f365be
fix(core): ensure compaction is more reliable, add reserve token buffer to ensure that input window has enough room to compact ( #12924 )
...
Co-authored-by: James Lal <james@littlebearlabs.io >
2026-02-10 19:55:22 -06:00
337c05849b
fix: use 'max' instead of 'xhigh' as highest reasoning effort for Anthropic models on Kilo Gateway ( #213 )
...
* 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
2026-02-10 11:47:14 -03:00
025dbcb4df
fix: use object-based reasoning format for Codex models with Kilo provider ( #206 )
...
Co-authored-by: kiloconnect[bot] <240665456+kiloconnect[bot]@users.noreply.github.com>
2026-02-10 10:41:12 +00:00
dd7da05bd9
feat: merge opencode v1.1.50
2026-02-04 18:24:00 -03:00
452320af42
feat: merge opencode v1.1.49
2026-02-04 17:06:38 -03:00
1a0b0077dc
feat: merge opencode v1.1.48
2026-02-04 14:48:20 -03:00
452ab52d47
feat: enable reasoning for Claude models via Kilo Gateway ( #101 )
...
* 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
2026-02-04 12:19:02 +01:00
3741516fe3
fix: handle nested array items for Gemini schema validation ( #11952 )
2026-02-03 09:24:52 -06:00
f27c1a8277
Revert "feat: add Claude extended thinking variants for Kilo Gateway ( #86 )" ( #87 )
...
This reverts commit 38c6b26db9
2026-02-03 10:22:12 +01:00
38c6b26db9
feat: add Claude extended thinking variants for Kilo Gateway ( #86 )
...
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%
2026-02-03 09:42:20 +01:00
3adeed8f97
fix(provider): strip properties/required from non-object types in Gemini schema ( #11888 )
2026-02-02 23:13:24 -06:00
1b5ef0737f
Merge OpenCode v1.1.42 ( #67 )
...
* chore: details
* fix(tui): add visual feedback for diff wrap and conceal toggles (#10655 )
* fix: await SessionRevert.cleanup for shell (#10669 )
* fix: Make diff wrapping toggle always available in command_list and correct a type error (resolves #10682 ) (#10683 )
* feat(app): default servers on web
* fix(app): handle non-tool call permissions
* fix(app): line selection waits on ready
* wip(app): model settings
* feat(app): model settings
* chore: generate
* wip(app): sidebar hover full
* wip(app): full-height sidebar
* wip(app): full-height sidebar
* feat(app): better sidebar hover when collapsed
* fix(app): query selector with non-latin chars
* chore: stale-while-revalidate headers for changelog
* chore: generate
* feat(opencode): add agent description (#10680 )
* feat: dynamically resolve AGENTS.md files from subdirectories as agent explores them (#10678 )
* chore: gen changelog page off changelog json
* upgrade opentui to v0.1.75
* chore: update nix node_modules hashes
* fix: move changelog footer outside content div to fix padding (#10712 )
* fix(app): Order themes alphabetically (#10698 )
* fix(app): restore external link opening in system browser (#10697 )
* fix(ui): use focus-visible instead of focus to prevent sticky hover effect on click (#10651 )
* fix: add 44px top padding to sticky version header on changelog (#10715 )
* feat(app): file tree
* wip(app): file tree mode
* wip(app): file tree mode
* chore: cleanup
* chore: cleanup
* chore(app): createStore over signals
* fix(enterprise): add message navigation to share page desktop view (#10071 )
Co-authored-by: Claude <noreply@anthropic.com >
* chore: fix changelog page
* Revert "fix(app): restore external link opening in system browser (#10697 )"
This reverts commit 984518b1c0#10590 )
* chore: generate
* test(app): e2e test for sidebar nav
* chore: refactor changelog
* fix(app): session load cap
* chore(app): missing i18n strings
* chore: cleanup
* fix(app): session sync issue
* chore: generate
* fix(ui): update button styles and disconnect button size
* fix(app): set provider row height to 56px
* fix(app): add hover text for env-connected providers
* fix(app): use default cursor for env provider text
* feat(ui): add providers icon and use in settings
* feat(ui): add models icon and use in settings
* fix(app): style view all button with interactive color and margin
* feat(ui): add optional transition animations to dialog
* fix(ui): prevent double-close and fix dialog replacement
* feat(app): add transition to select provider dialog
* feat(app): add transition to command palette
* fix(ui): reduce dialog transition in time to 150ms
* feat(app): add provider descriptions to settings
* feat(app): add Vercel AI Gateway provider description
* fix(ui): scope filetree pill tabs styling
* fix(app): adjust filetree panel padding
* fix(app): tighten filetree row spacing
* fix(app): use chevron icons for filetree folders
* fix(app): use medium font for filetree items
* fix(app): refine filetree row spacing and indent
* fix(app): reduce filetree folder indent
* fix(app): dim non-deep filetree guide lines
* fix(app): fade filetree guide lines on hover
* fix(app): move file tree toggle to right
* fix(app): disable tooltips in filetree tabs
* test(app): fix e2e test
* ci: add stale pr job
* ci: add dry-run option to stale PR closer workflow
Allows testing stale PR closure without actually closing PRs
* tweak: retry logic to catch certain provider problems
* fix(app): disconnect zen provider
* fix(app): zen disconnect not working
* chore: generate
* chore: cleanup
* fix: search clear icon
* wip: new release modal
- highlight key updates or new features
- needs some transition love
- all copy including text and video placeholder
* chore: cleanup
* wip: highlights
* wip: highlights
* wip: highlights
* tweak: when using messages api for copilot, attach anthropic beta headers
* chore: don't flip github draft release automatically
* fix(app): file tree performance
* fix(app): reactive file tree
* chore: generate
* chore: cleanup
* fix(app): enable file watcher
* chore: cleanup
* chore: cleanup
* test(app): file tree spec
* feat(release): add highlights template to draft releases (#10745 )
* docs: add Italian README (#10732 )
* chore: generate
* fix(app): add connect provier in model selector (#10706 )
* fix(desktop): disable magnification gestures on macOS (#10605 )
* feat: add /learn command to extract session learnings to scoped AGENTS.md files (#10717 )
* fix: use Instance.directory instead of process.cwd() in read tool
* fix(core): expose Instance.directory to custom tools
* fix(app): only show files in select dialog when clicking + tab
* desktop: deduplicate tauri configs
* ignore: update download stats 2026-01-27
* fix(app): auto-scroll button sometimes sticks
* chore(app): missing tooltips
* feat(app): forward and back buttons
* chore: cleanup
* feat(core): add worktree to plugin tool context
* fix(app): agent fallback colors
* fix(app): non-fatal error handling
* docs(core): plugin tool context updates
* fix(ui): set filetree tablist height to 48px with centered content
* fix(app): center filetree empty state with 32px top margin
* fix(app): update review empty states to 14px and align select file empty state
* fix(app): only show left border on plus button when sticky
* fix(app): use smaller close icon on tabs to match comment cards
* fix(ui): add max-width 280px to tabs with text truncation
* fix(app): don't show session skeleton after workspace reset
* perf(app): don't keep parts in memory
* perf(app): cleanup connect provider timers
* perf(app): shared terminal ghostty-web instance
* fix(app): select model anchor
* Revert "fix(app): select model anchor"
This reverts commit 36babf5fc390fd9a892ad3742690740296d94a23.
* fix(app): open markdown links in external browser
* fix(app): navigate to tabs when opening file
* chore(app): cleanup tailwind vs pure css
* fix(app): tooltips causing getComputedStyle errors in model select
* fix(app): swallow file search errors
* feat(opencode): Handle Venice cache creation tokens (#10735 )
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com >
* chore: generate
* chore(docs): Better explanation on how to allow tools in external directories (#10862 )
* tweak: adjust retry check to be more defensive
* ci: make tests passing a requirement pre-release
* zen: kimi k2.5 and minimax m2.1
* feat(app): update settings in general settings
* chore(app): translations
* test(app): new e2e smoke tests
* feat(app): show loaded agents.md files
* fix(app): session diffs not always loading
* feat(app): improved layout
* chore: generate
* tweak: add recommended topP/temp for kimi k2.5
* Set temperature for kimi k2.5
* Revert "ci: make tests passing a requirement pre-release"
This reverts commit 8c00818108bb63d16fa8#10887 )
* fix(app): auto-scroll
* feat: support headless authentication for chatgpt/codex (#10890 )
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com >
Co-authored-by: Aiden Cline <aidenpcline@gmail.com >
* fix(app): reintroduce review tab
* release: v1.1.37
* feat(app): use opentui markdown component behind experimental flag (#10900 )
* chore: generate
* docs: add Daytona OpenCode plugin to ecosystem (#10917 )
* chore: generate
* fix(cli): restore brand integrity of CLI wordmark (#10912 )
* zen: kimi k2.5
* fix(opencode): ensure unsub(PartUpdated) is always called in TaskTool (#9992 )
* chore: generate
* ignore: adjust flag
* bump plugin version
* ci
* fix: ensure parallel tool calls dont double load AGENTS.md
* release: v1.1.38
* fix: ensure copilot plugin properly sets headers for new messages api
* chore: generate
* release: v1.1.39
* bump: plugins
* ignore: update download stats 2026-01-28
* chore(app): global config changes
* feat(app): custom provider
* chore: cleanup
* wip(app): global config
* chore: cleanup
* chore: cleanup
* chore: generate
* feat(i18n): add th locale support (#10809 )
* chore: generate
* fix(desktop): enable ctrl+n and ctrl+p for popover navigation (#10777 )
* fix: add SubtaskPart with metadata reference (#10990 )
Co-authored-by: yangtianzhe <yangtianzhe@corp.netease.com >
* chore: generate
* fix(app): close review pane
* fix(markdown): Add streaming prop to markdown element (#11025 )
* release: v1.1.40
* fix(script): remove highlights template from release notes (#11052 )
* chore: generate
* fix(ui): allow KaTeX inline math to be followed by punctuation (#11033 )
* fix(app): alignment and padding in dialogs (#10866 )
* fix(app): fill missing zh keys to avoid English fallback (#10841 )
* chore: generate
* fix(opencode): normalize zh punctuation for Chinese UI (#10842 )
* fix(ui): improve zh duration display formatting (#10844 )
* fix(opencode): add `AbortSignal` support to `Ripgrep.files()` and `GlobTool` (#10833 )
* fix(typescript errors): remove duplicate keys causing typescript failures (#11071 )
* fix(app): file tabs - auto scroll on open & scroll via mouse wheel (#11070 )
* tweak: add ctx.abort to grep tool
* fix(app): types
* docs: update experimental environment variables in CLI docs (#11030 )
* fix(app): make settings more responsive for mobile and small web/desktop (#10775 )
* chore: generate
* fix(opencode): allow media-src data: URL for small audio files (#11082 )
* fix(provider): include providerID in SDK cache key (#11020 )
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com >
* fix: maxOutputTokens was accidentally hardcoded to undefined (#10995 )
* fix: ensure variants for copilot models work w/ maxTokens being set
* release: v1.1.41
* test: fix test
* chore: generate
* feat(cli): include cache tokens in stats (#10582 )
* ci: adjust team
* guard destroyed input field in timeout
* feat: Sequential numbering for forked session titles (Issue #10105 ) (#10321 )
* more timeout race guards
* fix: ensure that kimi doesnt have fake variants available
* fix: frontmatter was adding newlines in some cases causing invalid model ids (#11095 )
Co-authored-by: aptdnfapt <aptdnfapt@users.noreply.github.com >
* fix: undo change that used anthropic messages endpoint for anthropic models on copilot due to ratelimiting issues, go back to completions endpoint instead
* release: v1.1.42
* fix: fix types
* fix: fix kilocode/sdk type
* chore: update nix node_modules hashes
---------
Co-authored-by: adamelmore <2363879+adamdottv@users.noreply.github.com >
Co-authored-by: Idris Gadi <85882535+IdrisGit@users.noreply.github.com >
Co-authored-by: Noam Bressler <noamzbr@gmail.com >
Co-authored-by: Ariane Emory <97994360+ariane-emory@users.noreply.github.com >
Co-authored-by: GitHub Action <action@github.com >
Co-authored-by: Steffen Deusch <steffen@deusch.me >
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com >
Co-authored-by: Sebastian Herrlinger <hasta84@gmail.com >
Co-authored-by: Ryan Vogel <ryan@mandarin3d.com >
Co-authored-by: Sam Huckaby <samhuckaby@gmail.com >
Co-authored-by: Alex Yaroshuk <34632190+alexyaroshuk@users.noreply.github.com >
Co-authored-by: Rahul A Mistry <149420892+ProdigyRahul@users.noreply.github.com >
Co-authored-by: Adam <2363879+adamdotdevin@users.noreply.github.com >
Co-authored-by: zerone0x <hi@trine.dev >
Co-authored-by: Claude <noreply@anthropic.com >
Co-authored-by: Aiden Cline <aidenpcline@gmail.com >
Co-authored-by: Tim <hello@timsmart.co >
Co-authored-by: David Hill <iamdavidhill@gmail.com >
Co-authored-by: Fabio Martino <fabiomartino@users.noreply.github.com >
Co-authored-by: Chris Yang <18487241+ysm-dev@users.noreply.github.com >
Co-authored-by: Brendan Allan <git@brendonovich.dev >
Co-authored-by: dpuyosa <dpuyosa@users.noreply.github.com >
Co-authored-by: OpeOginni <107570612+OpeOginni@users.noreply.github.com >
Co-authored-by: Frank <frank@anoma.ly >
Co-authored-by: Rohan Godha <git@rohangodha.com >
Co-authored-by: opencode <opencode@sst.dev >
Co-authored-by: Tommy D. Rossi <beats.by.morse@gmail.com >
Co-authored-by: James Murdza <james@jamesmurdza.com >
Co-authored-by: Tito <github@mynameistito.com >
Co-authored-by: Goni Zahavy <goni1993@gmail.com >
Co-authored-by: Dax Raad <d@ironbay.co >
Co-authored-by: Nattawee Phantawong <crozzio@gmail.com >
Co-authored-by: Sairaj <sairaj2119@gmail.com >
Co-authored-by: RenegadeE <37529469+YangTianz@users.noreply.github.com >
Co-authored-by: yangtianzhe <yangtianzhe@corp.netease.com >
Co-authored-by: Max Kong <kongzheyuan@outlook.com >
Co-authored-by: Julian Coy <julian@juliancoy.com >
Co-authored-by: Babou <42550761+bbabou@users.noreply.github.com >
Co-authored-by: Devin Griffin <31415269+DNGriffin@users.noreply.github.com >
Co-authored-by: tan <zerotanapok@gmail.com >
Co-authored-by: Saba Tchikhinashvili <38084512+saba-ch@users.noreply.github.com >
Co-authored-by: ideallove <20013535+ideallove@users.noreply.github.com >
Co-authored-by: Benjamin Oldenburg <benjamin.oldenburg@ordis.co.th >
Co-authored-by: aptdnfapt <aptdnfapt@users.noreply.github.com >
Co-authored-by: marius-kilocode <marius@kilocode.ai >
2026-02-02 12:47:09 +01:00
f6948d0ffa
fix: variant logic for anthropic models through openai compat endpoint ( #11665 )
2026-02-01 14:50:49 -06:00
ca5e85d6ea
fix: prompt caching for opus on bedrock ( #11664 )
2026-02-01 14:42:42 -06:00
644f0d4e92
chore: generate
2026-01-31 02:35:22 +00:00
d9f18e4006
feat(opencode): add copilot specific provider to properly handle copilot reasoning tokens ( #8900 )
...
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com >
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com >
Co-authored-by: Aiden Cline <aidenpcline@gmail.com >
2026-01-30 19:53:22 -06:00
252b2c450d
chore: generate
2026-01-31 01:32:46 +00:00
0c32afbc35
fix(provider): use snake_case for thinking param with OpenAI-compatible APIs ( #10109 )
2026-01-31 01:32:02 +00:00
4a56491e42
fix(provider): exclude chat models from textVerbosity setting ( #11363 )
2026-01-30 21:06:48 +00:00
b938e1ea99
test: fix test
2026-01-28 22:02:14 -06:00
cf7e10c4e8
fix: add xhigh reasoning effort for GitHub Copilot GPT-5 models ( #10092 )
...
Co-authored-by: Arthur Freitas Ramos <arthur@MacBook-Air-de-Arthur.local >
2026-01-24 00:09:47 -05:00
0e1a8a1839
fix: Claude w/bedrock custom inference profile - caching support ( #9838 )
...
Co-authored-by: Aiden Cline <63023139+rekram1-node@users.noreply.github.com >
2026-01-21 10:42:13 -06:00
0d8e706fac
test: fix transfomr test
2026-01-18 14:44:39 -06:00
ea13b6e8aa
test: add azure test case
2026-01-17 00:35:49 -06:00
7c3eeeb0fa
fix: gpt id stuff fr fr this time :/ ( #9006 )
2026-01-16 22:09:36 -06:00
40836e9683
fix: fix the itemId stripping logic, this time it should fix that id issue w/ gpt models fr
2026-01-16 11:34:45 -06:00
d8ef9f808d
test: fix transform test
2026-01-15 19:27:04 -06:00
92931437c4
fix: codex id issue ( #8605 )
2026-01-15 01:31:50 -06:00
c285304acf
fix: for anthropic compat ensure empty msgs and empty reasoning is filtered out
2026-01-05 01:40:15 -06:00
2fd97377f6
test: fix transform test
2026-01-02 12:38:44 -06:00
81fef60266
fix: ensure variants also work for completely custom models ( #6481 )
...
Co-authored-by: Daniel Smolsky <dannysmo@gmail.com >
2025-12-30 16:37:32 -06:00
d89b567b47
fix: add transform case for gemini if mcp tool has missing array items ( #5846 )
2025-12-20 11:41:52 -06:00
edfe2e4f1c
test: fix test case
2025-12-19 16:37:36 -06:00
f9d0850c5e
test: add regression test for setCacheKey option
2025-12-17 10:24:53 -06:00
9eefcd1b41
Provider fix, anthropic Errorhandling if empty image file is read ( #5521 )
2025-12-14 23:56:47 -06:00
79a4c65313
fix: test
2025-12-14 21:42:17 -06:00
da56319af4
ignore: fix test
2025-12-10 13:31:45 -06:00
df64612d54
better interleaved thinking support ( #5298 )
2025-12-09 16:32:12 -06:00
46790e57e9
feat: Enhance DeepSeek reasoning content handling ( #4975 )
...
Co-authored-by: Aiden Cline <aidenpcline@gmail.com >
2025-12-03 21:31:36 -06:00
0c51feb9c2
fix: max tokens when using models like opus with providers other than anthropic ( #4307 )
2025-11-13 16:39:09 -06:00
Marius