100 Commits
Author SHA1 Message Date
Vegard Stikbakke 233f174401 fix(ai): use color Pi logo on OAuth pages 2026-10-01 15:10:11 +02:00
Vegard Stikbakke 0f8740bb65 fix(coding-agent): rename unconfigured to not configured 2026-10-01 12:37:37 +02:00
Vegard Stikbakke 8bdcd4498a fix(coding-agent): compact oversized trailing tool results
Fall back to the latest safe assistant boundary when trailing tool results exceed the retained-token budget, allowing older history to compact before the next provider request.

Closes #9740
2026-09-18 13:38:30 +02:00
Vegard Stikbakke bb0f4aa602 fix(ai): preserve DeepSeek V4 effort metadata
closes #9485
2026-09-18 10:43:20 +02:00
Vegard Stikbakke 8a7b0c03df fix(ai): price Bedrock one-hour cache writes
closes #9457
2026-09-15 14:18:15 +02:00
Vegard Stikbakke ceea48f5d5 Revert "fix(ci): Fix a broken test"
This reverts commit 71dca871bc.
2026-09-14 10:28:54 +02:00
Vegard Stikbakke 0c7bb7c5c7 fix(ai): identify Responses error providers
closes #9298
2026-09-14 10:14:48 +02:00
Vegard Stikbakke 713bdf38d5 Revert "fix(ai): update stale DeepSeek Flash test model IDs"
This reverts commit e8413008cd.
2026-09-11 15:08:41 +02:00
Vegard Stikbakke d92eb8d4b1 feat(ai): enable Fireworks Messages deferred tool loading 2026-09-10 21:18:35 +02:00
Vegard Stikbakke 519184eb65 fix(ai): accept Fireworks models in adaptive thinking metadata test
Update the catalog-wide assertion for metadata-driven Fireworks support while retaining the model-name restrictions for other providers. Add representative Fireworks models to the expected subset.
2026-09-10 10:31:35 +02:00
Vegard Stikbakke 6b94ae2ece fix(ai): preserve Fireworks thinking and native effort levels
Enable unsigned thinking replay and metadata-driven adaptive effort for Fireworks Messages models, with verified fallbacks for incomplete catalog metadata. Remove redundant GLM 5.2 and Kimi K3 effort aliases and add regression tests.

closes #9323
2026-09-10 09:51:31 +02:00
Vegard Stikbakke e86102f18f fix(ai): send Codex Off reasoning effort
closes #9191
2026-09-10 09:06:27 +02:00
Vegard Stikbakke 46bde88a1c feat(coding-agent): support per-model compaction token budgets
Closes #8133
2026-09-10 08:37:20 +02:00
Vegard Stikbakke f53ac11351 fix(coding-agent): collapse empty fullscreen footers
closes #8919
2026-09-08 14:16:18 +02:00
Vegard Stikbakke c1d4c80111 fix(coding-agent): embed all session status spinners in the editor border 2026-09-07 15:31:00 +02:00
Vegard Stikbakke 47acd8e6cf fix(coding-agent): preserve active operation UI during tree navigation
Follow-up to e687434a6.

Recheck compaction state after navigation dialogs and response cancellation before replacing status indicators or escape handlers. Document navigation conflicts and add regression coverage.
2026-09-07 12:59:20 +02:00
Vegard Stikbakke e44d75c20a fix(coding-agent): raise branch summary output cap
closes #8845
2026-09-03 15:14:51 +02:00
Vegard Stikbakke 007c0be640 docs(tui): clarify standalone renderer configuration
Follow-up to c505f4c19.
2026-09-03 12:03:59 +02:00
Vegard Stikbakke 23842b1e69 fix(coding-agent): tunnel proxied HTTP requests
closes #8134
2026-09-02 09:23:23 +02:00
Vegard Stikbakke b07e17faa2 fix(coding-agent): preserve partial Bash output when toggling thinking
closes #8611
2026-08-26 14:27:21 +02:00
Vegard Stikbakke 6c4f360264 fix(tui): chunk large main-screen renders closes #8028 2026-08-26 13:08:48 +02:00
Vegard Stikbakke 6b36eb5927 fix(coding-agent): remove explicit tool choice from compaction calls
Fixes #8649\n\nFixes #8638
2026-08-26 11:20:44 +02:00
Vegard Stikbakke 7af2d27dce fix(coding-agent): prevent Windows taskkill spawn crashes
closes #6596
2026-08-26 10:33:45 +02:00
Vegard Stikbakke a79b373342 feat(coding-agent): expose RPC queue clearing
closes #8432
2026-08-25 15:19:49 +02:00
Vegard Stikbakke fe37e9f9b5 fix(ai): omit tool_choice without tools closes #8607 2026-08-25 12:26:06 +02:00
Vegard Stikbakke 97fa14e39c fix(coding-agent): reject truncated compaction summaries closes #7048 2026-08-24 13:08:30 +02:00
Vegard Stikbakke 4495469a5e fix(coding-agent): compact without provider usage
closes #8328
2026-08-19 13:44:22 +02:00
Vegard Stikbakke 678f0af30d fix(coding-agent): show startup diagnostics in TUI
fixes #7829
2026-08-19 13:07:57 +02:00
Vegard Stikbakke 913bcf3391 fix(coding-agent): report settings diagnostic paths 2026-08-19 13:07:53 +02:00
Vegard Stikbakke 1e1a6e27be feat(coding-agent): include paths in settings errors 2026-08-19 13:07:20 +02:00
Vegard Stikbakke 3a0b9a3eee Revert "feat(agent): expose provider context construction"
This reverts commit 2509b5c037.
2026-08-19 09:53:15 +02:00
Vegard Stikbakke 8dab70281b Revert "feat(coding-agent): add cache-friendly compaction primitives"
This reverts commit cff1cf52c6.
2026-08-19 09:52:26 +02:00
Vegard Stikbakke cff1cf52c6 feat(coding-agent): add cache-friendly compaction primitives
Add provider-context prefix preparation and optional summary request settings needed by cache-friendly compaction. Existing AgentSession callers continue to pass undefined, so standalone summarization remains the default until the feature is wired in.
2026-08-18 14:52:06 +02:00
Vegard Stikbakke ef8dc7385b feat(coding-agent): centralize compaction summary requests 2026-08-18 14:19:01 +02:00
Vegard Stikbakke 836aee6d38 feat(coding-agent): show compaction usage notices 2026-08-18 13:27:00 +02:00
Vegard Stikbakke 2509b5c037 feat(agent): expose provider context construction 2026-08-18 09:34:47 +02:00
Vegard Stikbakke 8af7690c4f fix(coding-agent): skip trusted subagent prompts
closes #8261
2026-08-18 09:26:07 +02:00
Vegard Stikbakke 90305d90a0 fix(coding-agent): disable tools during summarization 2026-08-18 09:19:40 +02:00
Vegard Stikbakke e5dde9a76b feat(ai): add simple tool choice option 2026-08-18 08:41:24 +02:00
Vegard Stikbakke ad58801ce7 fix(ai): update Baseten GLM input modalities 2026-08-18 08:19:46 +02:00
Vegard Stikbakke 9117326b4c fix(ai): forward Azure Responses tool choice 2026-08-18 07:56:51 +02:00
Vegard Stikbakke 0e4d495414 fix(ai,coding-agent): remove deprecated Xiaomi models
closes #8187
2026-08-17 14:43:48 +02:00
Vegard Stikbakke af2c352238 fix(ai): honor Google thinking level maps
closes #8135
2026-08-17 13:10:55 +02:00
Vegard Stikbakke c7c763f5c4 fix(coding-agent): clarify truncated recovery failure
closes #8130
2026-08-17 11:03:07 +02:00
Vegard Stikbakke 58302d34e7 feat(coding-agent): support compaction routing sessions 2026-08-17 07:34:52 +02:00
Vegard Stikbakke 47bf47f11f docs(coding-agent): clarify compaction paths 2026-08-17 07:10:50 +02:00
Vegard Stikbakke 086c32e745 fix(ai): retry Copilot GET /models once on 429 during login
The login-time policy updates can drain the Copilot API rate-limit
bucket, in which case the follow-up GET /models is rejected with 429
(observed response: retry-after: 2, body 'too many requests') and login
aborts. Honor Retry-After and retry once instead of failing (#8121).
2026-08-15 07:33:05 +02:00
Vegard Stikbakke d5278eaac3 fix(ai): enable Copilot model policies sequentially during login
Batching policy POSTs at concurrency 4 still bursts 30 requests during
login, which trips the Copilot API rate limiter and makes the following
GET /models fail with 429, aborting login (#8121). Send the policy
updates one at a time instead.
2026-08-15 07:24:34 +02:00
Vegard Stikbakke 0d868f034e docs(agent): document repository lifecycle benchmarks 2026-08-14 12:56:36 +02:00
Vegard Stikbakke 0bddd620d6 feat(agent): benchmark session repository lifecycle 2026-08-14 12:56:36 +02:00
Vegard Stikbakke 4a92eb61a7 docs(agent): document session repository benchmarks 2026-08-14 12:39:18 +02:00
Vegard Stikbakke 0537b2f702 feat(agent): benchmark session repository forks 2026-08-14 12:38:11 +02:00
Vegard Stikbakke 48e4ef1f5d feat(agent): share session benchmark infrastructure 2026-08-14 12:17:40 +02:00
Vegard Stikbakke 7b8ee7a3f5 feat(agent): add session repository catalog benchmark 2026-08-14 12:17:40 +02:00
Vegard Stikbakke b863f1cff6 fix: revert SQLite conformance CI isolation 2026-08-14 11:16:49 +02:00
Vegard Stikbakke f7a00fe43b fix: isolate incomplete SQLite conformance job 2026-08-14 11:12:20 +02:00
Vegard Stikbakke a2dd8ab2a7 fix: skip incomplete SQLite conformance cases 2026-08-14 11:09:48 +02:00
Vegard Stikbakke 75bb6a0fcd feat(agent): add SQLite storage conformance coverage 2026-08-14 10:50:23 +02:00
Vegard Stikbakke 893d82525f feat(agent): isolate JSONL repository conformance tests 2026-08-14 10:48:45 +02:00
Vegard Stikbakke 21dec988de feat(agent): isolate JSONL storage conformance tests 2026-08-14 10:44:50 +02:00
Vegard Stikbakke 18ddf8ad90 fix(agent): test session destination reservation order 2026-08-14 10:24:04 +02:00
Vegard Stikbakke e3234b0fcb fix(agent): cover register rollback in storage conformance 2026-08-14 09:20:20 +02:00
Vegard Stikbakke 6f707eb360 fix(coding-agent): show managed-tool startup status in TUI 2026-08-13 15:54:34 +02:00
Vegard Stikbakke 46bb9a2c3b docs(coding-agent): clarify Windows paths in settings 2026-08-13 09:54:58 +02:00
Vegard Stikbakke 6d520c58d0 chore(docs): fix ascii alignment in compaction docs 2026-08-12 14:56:26 +02:00
Vegard Stikbakke b3edf01702 fix(ai): bound Copilot policy update concurrency 2026-08-11 14:25:53 +02:00
Vegard Stikbakke 452923b54a fix(tui): parse multiline LaTeX arguments
Closes #7760
2026-08-11 12:37:45 +02:00
Vegard Stikbakke 9cb7f493e7 docs(coding-agent): document AI_AGENT process marker
closes #7747
2026-08-11 09:04:09 +02:00
Vegard Stikbakke 31b513e316 chore(docs): change a bullet point in compaction docs 2026-08-10 08:27:35 +02:00
Vegard Stikbakke 4761021709 chore(docs): change 'reloads' phrasing in compaction docs 2026-08-10 08:03:46 +02:00
Vegard Stikbakke 368e013dec chore(docs): fix ascii alignment in compaction docs 2026-08-08 12:22:47 +02:00
Vegard Stikbakke d1a305613e feat(agent): add direct harness event listeners
Part of I2, "passive events and watch buffering," in packages/agent/docs/harness-v2.md.
2026-08-07 15:20:00 +02:00
Vegard Stikbakke 1dd2354052 feat(agent): add buffered harness event watches
This is incremental work towards I2 in the harness-v2.md plan.
2026-08-07 14:50:35 +02:00
Vegard Stikbakke 14ad9801be feat(agent): add harness events subscription interface 2026-08-07 13:21:36 +02:00
Vegard Stikbakke 4bf1bba203 docs(agent): reserve I2 2026-08-07 12:32:20 +02:00
Vegard Stikbakke e0900a6eaf docs(agent): clarify record query semantics 2026-08-07 11:01:26 +02:00
Vegard Stikbakke 87644c7b57 docs(agent): clarify branch query start requirement 2026-08-07 10:18:48 +02:00
Vegard Stikbakke f562a1a44f docs(agent): clarify branch query helpers 2026-08-07 09:55:13 +02:00
Vegard Stikbakke 80ef7ff0f4 fix(agent): populate session types with doc comments from design doc 2026-08-07 09:47:29 +02:00
Vegard Stikbakke 1026679525 docs(agent): unreserve R3 2026-08-07 08:56:05 +02:00
Vegard Stikbakke 69bab864d5 docs(agent): reserve harness v2 R3 2026-08-06 09:14:56 +02:00
Vegard Stikbakke c0947e6448 fix(ai): remove stale adaptive thinking model expectation
models.dev was updated and no longer includes the Vercel AI Gateway claude-opus-5-fast model.
2026-08-06 08:17:18 +02:00
Vegard Stikbakke abac29f671 docs(agent): reserve harness R2 work package 2026-08-05 15:28:37 +02:00
Vegard Stikbakke c518004666 docs(agent): clarify lane reduction contract (#7662)
The harness v2 plan is a bit underspecified for the state needed in R1. LaneState was missing the lane's config (model, thinking, tools). And the LaneReductionResult was missing, which is needed so we have failure information. Here's a suggestion at resolving.
2026-08-05 15:16:20 +02:00
Vegard Stikbakke ea7bfdc08b docs(agent): unreserve R2 work package 2026-08-05 14:03:24 +02:00
Vegard Stikbakke a5953d2e15 feat(agent): validate harness recovery record logs 2026-08-05 12:51:54 +02:00
Vegard Stikbakke afdf73661b docs(agent): reserve harness R1 work package 2026-08-05 12:12:31 +02:00
Vegard Stikbakke 1e95e16b61 fix(agent): remove redundant SQLite read clones 2026-08-05 10:51:46 +02:00
Vegard Stikbakke 591f22a618 feat(agent): add indexed harness recovery queries (#7646)
Implements R0 of the harness v2 plan, which adds RecordQuery.operationKind filtering and adds Session.findOpenOperations() for recovery purposes. The in-memory implementation keeps the open-operation projections in memory. SQLite implementation adds an index idx_records_session_lane_run_id_type.
2026-08-05 10:41:44 +02:00
Vegard Stikbakke f909da2bf0 docs(agent): reserve harness R0 work package 2026-08-05 10:11:40 +02:00
Vegard Stikbakke 97f0ccdd96 fix(coding-agent): recursively merge nested settings, closes #7572 2026-08-04 13:04:02 +02:00
Vegard Stikbakke d3da2e968a fix(ai): test composite OAuth refresh cancellation 2026-08-04 13:01:11 +02:00
Vegard Stikbakke 720f0e8eeb fix(ai): route Copilot Grok 4.5 through Responses, closes #7560 2026-08-04 12:42:01 +02:00
Vegard Stikbakke e741cb05ca fix(coding-agent): preserve extension auth endpoints
closes #7579
2026-08-04 12:37:02 +02:00
Vegard Stikbakke 42a06f9477 fix(coding-agent): test composite OAuth cancellation signal 2026-08-04 12:36:04 +02:00
Vegard Stikbakke b784c80961 fix(ai): remove OAuth refresh lock test 2026-08-04 12:19:19 +02:00
Vegard Stikbakke acbdc0d25e fix(ai): bound OAuth refresh duration
closes #7508
2026-08-04 12:12:32 +02:00
Vegard Stikbakke ab5f8d88ee fix(coding-agent): retain heterogeneous auth model types
closes #6768
2026-08-04 11:05:27 +02:00
Vegard Stikbakke 18d65de623 fix(coding-agent): preserve Copilot summary endpoint
closes #6768
2026-08-04 11:00:30 +02:00
Vegard Stikbakke 0524d6897f fix(coding-agent): prevent flaky RPC prompt tests 2026-08-04 09:50:42 +02:00