agent-memory-governance

What problem does it solve?

This Skill governs persistent agent memory so it stays useful without becoming dangerous, preventing stale, duplicated, secret-leaking, or unverified facts from steering future sessions.

Core Features & Use Cases

• Memory Rule Setting: Defines what facts may be stored, including provenance, absolute dates, and one-fact-per-entry hygiene.
• Trust Verification: Checks remembered repo, PR, and branch state against live git and gh output before any action is based on memory.
• Memory Auditing and Cleanup: Classifies entries as verified-current, stale, wrong-when-written, duplicate, forbidden-content, or unverifiable, then proposes keep, correct, merge, or delete dispositions.
• Use Case: A prior session remembered that a PR was still open; this Skill verifies the live state, marks the entry stale if it was merged, and proposes a corrected replacement with provenance.

Quick Start

Use the agent-memory-governance skill to audit our memory store, verify recalled repo state against live git and gh output, and propose safe per-entry corrections without editing files yet.