Commit Graph
6490 Commits
Author SHA1 Message Date
Armin Ronacher 40c256cccb feat(coding-agent): defer extension loader dependencies
Load ordinary jiti lazily on Node while retaining jiti/static for Bun and SEA binaries. This closes #9540.
2026-09-19 23:25:47 +02:00
xu0o0 bfa6862400 fix(tui): handle CJK punctuation in file autocomplete (#9746) 2026-09-19 21:53:26 +02:00
Armin Ronacher c596d09d9c feat(coding-agent): add prompt cache warming (#9668) 2026-09-19 21:18:15 +02:00
Armin Ronacher 7d5eb0ee3b fix(ai): mock Radius in model generation tests 2026-09-19 19:41:20 +02:00
Armin Ronacher eba619879c docs: audit unreleased changelogs 2026-09-19 19:33:31 +02:00
Armin Ronacher 4d38031fbd feat(ai): ship Radius model catalog 2026-09-19 19:23:31 +02:00
Armin Ronacher de2de549bc fix(coding-agent): close compaction cancellation races
Closes #9340 and closes #9777.
2026-09-19 19:18:18 +02:00
Armin Ronacher 3c75b27479 feat(coding-agent): add bug reporting 2026-09-19 17:55:49 +02:00
Mario Zechner 36b60d2e89 fix: clean up delta test lint diagnostics 2026-09-19 00:26:08 +02:00
Mario Zechner 58541ee721 fix: reduce delta proxy retention and compact cloned objects
Keep clone and GC regressions plus a standalone traversal benchmark. Archive the discarded graph experiment and baseline comparisons in durable docs, including the remaining transient-memory and traversal-speed costs.
2026-09-19 00:25:44 +02:00
Mario Zechner e80cf14011 docs(durable): simplify document runtime contracts 2026-09-19 00:20:08 +02:00
Mario Zechner 734ab3434f Merge remote-tracking branch 'origin/main' into HEAD 2026-09-18 22:18:01 +02:00
Mario Zechner 0db5659249 fix(durable): index memory storage queries 2026-09-18 22:17:36 +02:00
Mario Zechner a16ccd9be8 feat(durable): refine record and query contracts 2026-09-18 21:01:46 +02:00
David Brailovsky e4ce7b449f fix(ai): update Kimi model catalog source 2026-09-18 16:04:16 +02:00
Christian Klotz 59eb4c393c fix(coding-agent): clarify copy shortcut description (#9745)
Fixes #9660
2026-09-18 14:12:14 +01:00
David Brailovsky c1263aa113 fix: clean up delta lint diagnostics 2026-09-18 15:00:25 +02:00
David Brailovsky 1e39862f67 fix(coding-agent): detect llama.cpp chat-template thinking
Inspect loaded models through /props and configure boolean thinking levels for templates supporting enable_thinking.

Fixes #9528
2026-09-18 14:54:17 +02:00
Mario Zechner b5ef419d5b docs(agent): restore non-Pico5 documentation 2026-09-18 13:48:19 +02:00
Mario Zechner cf33309117 Merge remote-tracking branch 'origin/main' 2026-09-18 13:39:36 +02:00
Mario Zechner 0801601621 feat(durable): move Pico into dedicated package 2026-09-18 13:39:09 +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
David Brailovsky 5901446094 fix(tui): reduce fuzzy search latency
Use native substring search to find successive fuzzy-match characters while preserving scoring and result ordering.

Closes #9267
2026-09-18 12:47:58 +02:00
Mario Zechner eed5263cdd Merge remote-tracking branch 'origin/main' into HEAD 2026-09-18 12:33:14 +02:00
Mario Zechner cf8d5fac30 feat(agent): add Pico storage foundation 2026-09-18 12:32:00 +02:00
Ramiz Wachtler fe219d7f8d feat(coding-agent): format bash tool durations to support hours, minutes, seconds (#9742) 2026-09-18 12:25:15 +02:00
David Brailovsky 661619e872 fix(ai): scope bodyless overflow errors to Cerebras
closes #9482
2026-09-18 11:15:48 +02:00
Vegard Stikbakke bb0f4aa602 fix(ai): preserve DeepSeek V4 effort metadata
closes #9485
2026-09-18 10:43:20 +02:00
David Brailovsky 13784598d2 fix(coding-agent): suppress repeated Anthropic thinking drop notices
Fixes #9391
2026-09-18 09:58:42 +02:00
David Brailovsky ea9e093516 fix(ai): update tests for current model catalogs 2026-09-18 09:49:53 +02:00
badlogic 328926b30e docs(chord): describe tracker ownership and aliasing as they behave 2026-09-18 00:58:07 +00:00
badlogic c4289b20eb fix(chord): keep tracked paths correct across structural mutation
A wrapper baked its path at creation, so a reference held across a
structural mutation kept addressing its old index: unshift then a write
through a held element edited its neighbour, and a write after a splice
could produce a path the replica cannot resolve. The baseline diff this
tracker replaced was immune, because its dirty paths were only hints and
the ops came from comparing values.

A wrapper now holds a cell (parent link plus segment) and the path is
walked at use time, so renumbering an element corrects every descendant.
pop/shift/unshift/splice and length truncation shift live children by the
delta and mark removed ones dead; sort/reverse/fill/copyWithin relocate
them by identity. Renumbering iterates a set of live cells rather than the
lookup cache, which is keyed by index and can evict a child that is still
held. Writes through an element that left the document mutate it and
record nothing, as plain JS and the baseline diff both do.

One object can occupy several positions in the tree. A wrapper is now
shared per target and holds a cell per position, and a write emits an op
for each, which is what the baseline diff produced for the same shape.
Positions are registered when a path is read, when a tracked value is
assigned elsewhere, and when one is inserted by an array mutator (only the
inserted range is examined; scanning the array made every structural
mutation O(n)). A blocked view of an object gets its own wrapper so the
reserved-key guard cannot be bypassed through a safe alias.

Paths are cached per cell against a counter that only structural mutation
bumps, and the per-write cell walk is skipped until a document actually
aliases something.
2026-09-18 00:32:50 +00:00
Ramiz Wachtler 46c9de402b feat(coding-agent): add event handler unsubscribe (#9630) 2026-09-17 21:44:26 +02:00
Armin Ronacher 16292398af fix(coding-agent): send forced system prompts without recording them
Remove SystemMessage.replace. A prompt forced from before_agent_start is a rendering
of the current prompt for the run, not conversation state: persisting it as a
replacing system message wrote the full prompt and tool list on every change,
lost the structured sections while forced, and disabled mid-conversation system
messages for the rest of the session once a replacement existed.

The transcript now always records section patches. AgentSession wraps
transformContext and, when a forced prompt is active, collapses the system
messages into one head carrying the forced text and the current tools. Providers
receive the same request as before; nothing is persisted and the projection ends
with the run.

The evals harness validated the without_docs variant from the transcript, which
relied on the persisted replacement. Its transform extension now records the
prompt it forced and validates that instead.
2026-09-17 20:02:23 +02:00
David Brailovsky 7811394112 docs(coding-agent): update changelog 2026-09-17 17:51:25 +00:00
David Brailovsky a8b3dd1998 fix(coding-agent): fail signal-terminated shell commands
Map signal termination to conventional nonzero exit codes and reject null exit codes from custom shell operations while preserving partial output.

fixes #9577
2026-09-17 17:01:48 +02:00
David Brailovsky 3955b27a1d fix(ai): preserve Vercel AI Gateway unsigned thinking
Enable empty-signature thinking replay for all generated Vercel AI Gateway models.

closes #9676
2026-09-17 16:43:24 +02:00
David Brailovsky 4658534986 fix(coding-agent): shorten Anthropic thinking drop notices 2026-09-17 16:38:56 +02:00
Mario Zechner 2c995acf44 feat(chord): replace delta tracker with operation log 2026-09-17 16:36:37 +02:00
Mario Zechner 7e19507681 feat(coding-agent): add experimental micro agent 2026-09-17 16:33:03 +02:00
Mario Zechner 0e19ac1161 Merge remote-tracking branch 'origin/main' into pico 2026-09-17 16:31:25 +02:00
Christian Klotz 5a3a03a7f5 fix(coding-agent): validate eval prompts from transcripts (#9706)
Preserve partial-run usage and session artifacts when post-run validation fails.
2026-09-17 14:12:10 +01:00
Mario Zechner 729d5cb74d docs(agent): finalize Pico5 specification set 2026-09-17 14:57:03 +02:00
Christian Klotz 42cd371ba4 feat(coding-agent): add TUI context footer eval (#9705) 2026-09-17 13:21:31 +01:00
David Brailovsky e98f287ee4 fix(ai): retry Azure peak-load capacity errors
Closes #9669
2026-09-17 12:01:39 +02:00
Mario Zechner 2b04ce27fb docs(agent): refine Pico5 plugin state API 2026-09-17 11:24:54 +02:00
github-actions[bot] 7140838fdd chore: approve contributors from issue #9645 2026-09-17 09:24:25 +00:00
David Brailovsky e5d18382a2 fix(ai): retry Cloudflare 520 responses
closes #9627
2026-09-17 10:39:48 +02:00
David Brailovsky 1283afd0d0 fix(ai): preserve thinking replay through renamed Anthropic models
Keep the requested model ID on assistant messages and record the
provider-reported model separately, while retaining fallback pricing.

Fixes #9188
2026-09-17 10:26:03 +02:00
David Brailovsky 16235fd935 fix(ai): avoid unsupported Gemini thinking levels
Use each model's supported levels when disabling or selecting Gemini thinking.

Fixes #9455
2026-09-17 10:08:36 +02:00