This release includes 1 breaking change for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Summary
AI summaryRenamed compliance reason field and added search suggestion stopword filtering.
Full changelog
Attend Protocol Quality Improvements
Improved
- Compliance context-awareness for chatter turns.
turnsWithoutWriteno longer escalates on pure chatter turns (pre-response → post-response with no mid-turn activity). Only turns where the agent performed tool work count toward the write compliance threshold. - Search suggestion stopword filtering. Added
tokenizeForSearch()with ~80 stopwords. Conversational turns like "Give me possible ideas" no longer produce search suggestions with generic terms. - Attend reason clarity. Renamed
memory_needed_no_factstomemory_checked_no_match. AddedmemorySearchPerformedandmemoryResultsConsideredfields toAttendResultfor transparent search reporting. - Auto-capture observability. Post-response attend now surfaces
PostResponseCaptureInfoshowing how many facts were extracted, written, and skipped byrememberAssistantResponseFacts. - Semantic tag-aware retrieval.
observe()now accepts an optionalsemanticFilter(domains, intents, scopes, tags) and boosts matching facts during ranking viasemanticMatchScore().
Testing
- 5 new test files with 33 scenarios covering all 5 features
- All existing test suites pass with zero regressions
Breaking Changes
- Renamed `memory_needed_no_facts` to `memory_checked_no_match` in AttendResult.
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About nfemmanuel/iranti
Persistent shared memory for AI coding agents. Stores facts as `entity/key/value` triples with hybrid semantic search, task checkpoints, and conflict resolution — shared across Claude Code, Codex CLI, and GitHub Copilot.
Related context
Related tools
Beta — feedback welcome: [email protected]