This release adds 2 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Summary
AI summaryFixed attend() to recover personal-memory facts without aggressive memory_not_needed defaults.
Full changelog
0.2.13
Changed
- Hybrid search now falls back to deterministic in-process semantic scoring when pgvector is unavailable, instead of collapsing to lexical-only ranking.
- TypeScript and Python clients now default entity hints from IRANTI_MEMORY_ENTITY when callers do not provide explicit hints.
- Mock-provider behavior is now contract-faithful for structured extraction, entity detection, memory classification, and relevance filtering.
Fixed
- attend() no longer defaults ambiguous prompts to memory_not_needed so aggressively, and can now recover personal-memory facts like user/main without manual entity hints.
- Hybrid search no longer returns zero-signal lexical rows so readily, and external vector backends now receive the metadata needed for filtered searches.
- Python relationship documentation now matches the actual response shape, and the Python smoke test now covers relationship readback.
- The Python smoke script no longer crashes on Windows console encoding when maintenance errors contain non-ASCII characters.
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]