Commit Graph

15 Commits

Author SHA1 Message Date
Peter Steinberger
6da03eabe2 fix: add changelog and clean regression comment for tool-result guard (#25429) (thanks @mikaeldiakhate-cell) 2026-02-24 14:42:09 +00:00
Leakim
8db7ca8c02 fix: prevent synthetic toolResult for aborted/errored assistant messages
When an assistant message with toolCalls has stopReason 'aborted' or 'error',
the guard should not add those tool call IDs to the pending map. Creating
synthetic tool results for incomplete/aborted tool calls causes API 400 errors:
'unexpected tool_use_id found in tool_result blocks'

This aligns the WRITE path (session-tool-result-guard.ts) with the READ path
(session-transcript-repair.ts) which already skips aborted messages.

Fixes: orphaned tool_result causing session corruption

Tests added:
- does NOT create synthetic toolResult for aborted assistant messages
- does NOT create synthetic toolResult for errored assistant messages
2026-02-24 14:42:09 +00:00
Peter Steinberger
bfada9e425 test: move more local agents helper suites out of e2e 2026-02-22 10:55:22 +00:00
Peter Steinberger
9131b22a28 test: migrate suites to e2e coverage layout 2026-02-13 14:28:22 +00:00
Peter Steinberger
85409e401b fix: preserve inter-session input provenance (thanks @anbecker) 2026-02-13 02:02:01 +01:00
Tyler Yust
0deb8b0da1 fix: recover from context overflow caused by oversized tool results (#11579)
* fix: gracefully handle oversized tool results causing context overflow

When a subagent reads a very large file or gets a huge tool result (e.g.,
gh pr diff on a massive PR), it can exceed the model's context window in
a single prompt. Auto-compaction can't help because there's no older
history to compact — just one giant tool result.

This adds two layers of defense:

1. Pre-emptive: Hard cap on tool result size (400K chars ≈ 100K tokens)
   applied in the session tool result guard before persistence. This
   prevents extremely large tool results from being stored in full,
   regardless of model context window size.

2. Recovery: When context overflow is detected and compaction fails,
   scan session messages for oversized tool results relative to the
   model's actual context window (30% max share). If found, truncate
   them in the session via branching (creating a new branch with
   truncated content) and retry the prompt.

The truncation preserves the beginning of the content (most useful for
understanding what was read) and appends a notice explaining the
truncation and suggesting offset/limit parameters for targeted reads.

Includes comprehensive tests for:
- Text truncation with newline-boundary awareness
- Context-window-proportional size calculation
- In-memory message truncation
- Oversized detection heuristics
- Guard-level size capping during persistence

* fix: prep fixes for tool result truncation PR (#11579) (thanks @tyler6204)
2026-02-07 17:40:51 -08:00
Shakker
befa421a57 Agents: flush pending tool results on drop 2026-02-02 23:56:27 +00:00
Justin
0da6de6624 Agent: repair malformed tool calls and session files 2026-02-02 23:56:27 +00:00
cpojer
f06dd8df06 chore: Enable "experimentalSortImports" in Oxfmt and reformat all imorts. 2026-02-01 10:03:47 +09:00
Peter Steinberger
2424404fb4 docs: add transcript hygiene reference 2026-01-23 01:34:21 +00:00
Peter Steinberger
7fb45ed9b8 fix: strip final tags from session messages 2026-01-16 02:16:17 +00:00
Peter Steinberger
c379191f80 chore: migrate to oxlint and oxfmt
Co-authored-by: Christoph Nakazawa <christoph.pojer@gmail.com>
2026-01-14 15:02:19 +00:00
Peter Steinberger
3ffb9a3b5e fix: keep session sanitizer stable 2026-01-12 18:19:30 +00:00
Peter Steinberger
e91aa0657e fix: add copilot tests and lint fixes 2026-01-12 17:48:08 +00:00
Peter Steinberger
f5d5661adf fix: guard session tool results 2026-01-12 17:28:46 +00:00