This release includes breaking changes for platform teams planning a safe upgrade.
✓ No known CVEs patched in this version
Summary
AI summaryCopilot now uses iranti_search for recall questions, with imperative hint and updated instruction templates.
Full changelog
What's changed
Fix: Copilot (and other models) now use iranti_search for recall questions
Three connected improvements to stop models from reading the codebase when they should be querying Iranti memory:
-
searchSuggestion.hintis now imperative — wheniranti_attendfinds no auto-inject facts, the response now says "Call iranti_search BEFORE reading the codebase — empty attend facts do NOT mean data is absent" instead of the passive "Try iranti_search with..." that models ignored. -
copilot-instructions.mdtemplate gets a Recall section — explicitly instructs models to calliranti_searchbefore the codebase when the user asks about prior work/status, and to act onsearchSuggestionwhen attend returns it. -
copilot-instructions.mdtemplate gets a Writes section — instructs models to use specific entities (e.g.project/iranti_benchmarking) rather than the top-level project entity, so facts are stored where future attend injection will find them.
Run iranti copilot-setup (or iranti integrate copilot) in your project to pick up the updated instructions template.
No breaking changes. All 149 contracts pass.
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]