MeMesh Review

Health-scores and cleans up your MeMesh AI memory database

von PCIRCLE-AI · PCIRCLE-AI/memesh-llm-memory

Funktioniert mit Setup ★ 7.6/10

MeMesh Review — Health-scores and cleans up your MeMesh AI memory database

Was es kann

Runs the MeMesh CLI/MCP to pull recent memories, computes a health score (activity, quality, freshness, self-improvement), flags stale/verbose/conflicting entries, and proposes forget/consolidate actions. Triggers on 'review memories', 'check memory health', 'clean up knowledge', or 'what's in my memory' — but only useful if the separate MeMesh memory system (@pcircle/memesh) is already installed and populated.

Testbericht

Real run against a live memesh DB produced a clean structured health report, but the Step-1 commands `memesh recall --tag "type:decision"` etc. return empty arrays against the actual CLI (types are a field, not a `type:x` tag) — a literal-follow agent silently undercounts. Also points to `./scripts/verify-docs-sync.sh` (repo-root only, not in the skill folder) and an undefined external plugin `@sa:comprehensive-code-review`.

Getestet am: 2026-07-16 · Claude Code 2.x (agent harness)

Installation

git clone https://github.com/PCIRCLE-AI/memesh-llm-memory
mkdir -p ~/.claude/skills
cp -r memesh-llm-memory/skills/memesh-review ~/.claude/skills/memesh-review

Befehle & Beispiel-Prompts

/memesh-reviewHealth-scores and cleans up your MeMesh AI memory database

Skills reagieren auf normale Anfragen — keine Slash-Befehle nötig. Nach der Installation aktivieren Prompts wie diese den Skill (auf Englisch):

Check the health of my AI memory database and flag any stale entries
Clean up conflicting and overly verbose memories in my knowledge store
Review what's currently in my memory system and suggest what to forget

Vollständiger Testbericht (Englisch) →