Commit Graph

1010 Commits

Author SHA1 Message Date
David Brailovsky 162179af5a add branch/compact retries to agent-harness 2026-07-21 12:11:39 +00:00
David Brailovsky 8e53e0e49c compaction & branch summarization follow retry policy
fixes #6647

compaction (auto & manual) and branch summarization retry on transient failures.
use the same retry policy from settings.
emit events for the tui to show indication of retries
2026-07-21 10:07:20 +00:00
David Brailovsky 890b3547af run generate-models
fixes #6891
2026-07-21 07:48:27 +00:00
Armin Ronacher 3a40794ea1 feat(ai): regenerate models 2026-07-20 17:19:40 +02:00
David Brailovsky 2fd3868401 add usage info to branch summary, compaction and tool result entries (#6671)
* add usage info to branch summary entries

* add usage to compaction entries

* allow custom tools to report llm usage in tool results

* allow observing and patching usage in tool_result hooks

* agent-harness: save usage in entries

for compaction, branch summaries and tool results
2026-07-20 16:41:43 +02:00
Cristina Poncela Cubeiro 1942b2600f fix: env section ignored (#6864) 2026-07-20 16:20:49 +02:00
QuintinShaw bbb91fa8ae feat(ai): add Qwen Token Plan as built-in provider (#6858)
Add Alibaba Cloud Model Studio Token Plan subscription service as two
built-in API-key providers: qwen-token-plan (international, Singapore)
and qwen-token-plan-cn (China, Beijing).

Each provider exposes 15 text-generation models (Qwen, DeepSeek, GLM,
Kimi, MiniMax) via the OpenAI-compatible endpoint with DashScope
enable_thinking support. Model metadata is sourced from models.dev;
qwen3.8-max-preview is hardcoded until models.dev includes it.

Also fixes kimi-coding test references (k2p7 -> kimi-for-coding) after
models.dev catalog update picked up by generate-models.

Closes #6850
2026-07-20 13:53:30 +02:00
Cristina Poncela Cubeiro d9f7f81473 fix: tool_call_id error when switching (#6854) 2026-07-20 13:51:16 +02:00
David Brailovsky 21f92b319d run generate-models 2026-07-20 10:21:34 +00:00
David Brailovsky 8b93737078 kimi: add low,high to k3 and remove k2p7 references 2026-07-20 09:00:15 +00:00
Alexey Zaytsev 94373d815d feat(ai): add shared contentText utility (#6840)
Co-authored-by: Armin Ronacher <armin.ronacher@active-4.com>
2026-07-20 01:50:36 +02:00
Armin Ronacher 75cb0b873a fix(ai): support responses API for OpenCode Go 2026-07-20 01:28:57 +02:00
Alexey Zaytsev d2f8dafb0f fix(ai,agent,coding-agent): share UUIDv7 and use for Codex (#6834) 2026-07-20 01:23:47 +02:00
Armin Ronacher 956074697f fix(ai): replace generic record checks 2026-07-19 22:21:31 +02:00
David Brailovsky 5124c61b25 buildfix: fix tests referencing old models 2026-07-17 13:24:11 +00:00
Armin Ronacher a9f6a3159a feat(ai): separate generated model data (#6765) 2026-07-17 11:08:48 +02:00
David Brailovsky b0c2a90e51 fix(coding-agent): retry OpenAI Responses early EOF
closes #6727
2026-07-17 11:00:03 +02:00
Armin Ronacher 8881e17625 fix(ai): show implied Kimi Coding subscription costs 2026-07-17 10:01:22 +02:00
Armin Ronacher 8dc78834cd Release v0.80.10 2026-07-16 23:59:41 +02:00
Armin Ronacher b8575f60fa fix(ai): use adaptive thinking for Kimi Coding 2026-07-16 20:59:54 +02:00
Armin Ronacher aba324504c fix(ai): correct Kimi K3 Moonshot pricing 2026-07-16 20:37:48 +02:00
Armin Ronacher c1b7856e26 fix(ai): prevent removed xAI models regenerating, closes #6736 2026-07-16 20:07:36 +02:00
Armin Ronacher 78ff249436 fix(ai): expose max thinking for Kimi Coding K3
Closes #6737
2026-07-16 19:58:17 +02:00
Armin Ronacher 2d16f92973 Release v0.80.9 2026-07-16 19:17:57 +02:00
Milosz Jankiewicz a01baaaea7 feat(xai): prefilled OAuth device link, SuperGrok login label, trimmed model list (#6734)
- Use verification_uri_complete from the xAI device-code response so the
  login link opens with the user code prefilled (https-validated)
- Add OAuthAuth.loginLabel and show "Sign in with SuperGrok or X Premium"
  in the xAI auth-type selector (threaded through lazyOAuth)
- Remove grok-3, grok-3-fast, grok-4.20 variants and grok-code-fast-1
  from the built-in xAI catalog and default to grok-4.5

Co-authored-by: Jaaneek <Jaaneek@users.noreply.github.com>
2026-07-16 18:59:54 +02:00
Armin Ronacher c2c32febb0 fix(ai): correct Kimi K3 gateway output limits 2026-07-16 18:56:54 +02:00
David Brailovsky 70c5763297 add deferred tool loading for kimi-k3 2026-07-16 15:39:22 +00:00
David Brailovsky 3524cd4c85 Merge remote-tracking branch 'origin/main' into add-kimi-deferred-tools 2026-07-16 17:30:50 +02:00
David Brailovsky 6a884b4a75 add kimi-k3 2026-07-16 15:29:32 +00:00
David Brailovsky fae7176cb9 Release v0.80.8 2026-07-16 16:35:16 +02:00
Armin Ronacher 6442536b1e fix(coding-agent): bundle OAuth flows in Bun binaries 2026-07-16 15:36:16 +02:00
David Brailovsky f16b4e0cda add deferred tools support for kimi in openai-completions api 2026-07-16 12:26:56 +00:00
Armin Ronacher 97f9978fa6 feat(coding-agent): add model catalog refresh flag 2026-07-16 12:46:49 +02:00
Milosz Jankiewicz 5220aba619 feat(ai): add xAI device OAuth and route grok-4.5 through Responses (#6651)
* feat(ai): add xAI device OAuth and route grok-4.5 through Responses

Add xAI device-code OAuth alongside XAI_API_KEY. Route only grok-4.5
through Responses with low/medium/high reasoning; other xAI models stay
on Completions.

* fix(ai): tolerate xAI device-code interval 0 and correct token poll error label

---------

Co-authored-by: Jaaneek <Jaaneek@users.noreply.github.com>
2026-07-16 09:58:24 +02:00
Mario Zechner 5e336cfa80 Merge origin/main into model runtime changes 2026-07-15 13:00:45 +02:00
Mario Zechner bd9e09db44 feat(coding-agent): expose dynamic provider refresh 2026-07-15 12:56:50 +02:00
Mario Zechner cd7cad4ee2 feat(coding-agent): merge origin/main into model runtime facade 2026-07-15 12:25:36 +02:00
David Brailovsky dcfe36c797 clamp session-id to 64 chars for openai-codex (#6653)
fixes #6630
2026-07-14 21:09:29 +02:00
Armin Ronacher 818d67457c Release v0.80.7 2026-07-14 18:36:46 +02:00
Mario Zechner 9993c96907 feat(coding-agent): replace model registry with model runtime
Move provider auth and OAuth flows onto pi-ai Models, compose models.json and extension overlays through ModelRuntime, and retain ModelRegistry as an extension compatibility facade.
2026-07-14 17:48:45 +02:00
David Brailovsky 1f9e846c84 don't send session-id header to opencode openai-responses models (#6645)
fixes #6625
2026-07-14 15:57:01 +02:00
Armin Ronacher 961fa6c142 feat(ai): add Radius gateway support 2026-07-14 11:01:21 +02:00
David Brailovsky 0e6909f050 anthropic-messages: skip usage fields if empty (#6611)
fixes: #6567
2026-07-13 18:07:59 +02:00
David Brailovsky 1f0dbc008c backfill encrypted_content from response.completed for missing reasoning blocks (#6608)
fixes #6409
2026-07-13 17:27:36 +02:00
David Brailovsky f8f75544b6 pass bedrock unhandled stop reasons to error message (#6598)
fixes #6485
2026-07-13 11:39:28 +02:00
Petr Houška 298665cfb9 fix(ai): support OpenRouter session affinity (#6496)
Fixes #6366
2026-07-13 09:57:53 +02:00
Petr Houška f7b78e2a42 fix(ai): route GitHub Copilot MAI-Code models through /responses endpoint (#6544)
mai-code-1-flash-picker is not accessible via /chat/completions and
must use the Copilot /responses endpoint. Update generate-models.ts to
send mai-* models to the responses API and regenerate the affected
model entry.
2026-07-13 09:56:40 +02:00
Alexey Zaytsev eacaa130ab ai: OpenAI and Codex forced tool calls (#6588)
* feat(ai): support forced OpenAI Codex tool calls

* feat(ai): support OpenAI Responses tool choice
2026-07-13 09:48:33 +02:00
Mark Phelps bdd5c53bcd fix(ai): fall back to ambient Cloudflare account id for key-only credentials (#6292)
Cloudflare Workers AI / AI Gateway resolved provider config from the
credential only, never consulting ambient env for a field the credential
omitted. The coding-agent /login flow stores just the API key, so
CLOUDFLARE_ACCOUNT_ID lives only in the environment; the key-only credential
short-circuited the env lookup, the account id stayed unresolved, and requests
hit the literal {CLOUDFLARE_ACCOUNT_ID} base URL -> 404.

resolveValue now merges per field: prefer the credential value, fall back to
ctx.env(name).

closes #6021

Signed-off-by: Mark Phelps <209477+markphelps@users.noreply.github.com>
Co-authored-by: Mario Zechner <badlogicgames@gmail.com>
2026-07-11 14:30:40 +02:00
Armin Ronacher 850c210b77 fix(ai): filter ambient auth markers in compat dispatch 2026-07-11 13:38:48 +02:00