Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/PCIRCLE-AI/memeshnpx agentmods add skills/pcircle-ai/memesh/memesh-reviewWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/pcircle-ai/memesh/memesh-review)<a href="https://agentmods.dev/skills/pcircle-ai/memesh/memesh-review"><img src="https://agentmods.dev/badge/skills/pcircle-ai/memesh/memesh-review.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00064 | $0.00933 |
| Opus 5 | $0.00032 | $0.00466 |
| Sonnet 5 | $0.00013 | $0.00187 |
| Haiku 4.5 | $0.00006 | $0.00093 |
Grade A, and why
memesh-review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MeMesh Memory Review
Review the memory database and provide actionable cleanup recommendations.
How to Access
Use CLI (works everywhere) or MCP tools (if available). See the memesh skill for auto-detect instructions.
Process
Step 1: Gather data
# Get system health
memesh status
# Get all recent memories (structured output for analysis)
memesh recall --limit 50 --json
# Get memories by type for quality analysis
memesh recall --tag "type:decision" --json
memesh recall --tag "type:lesson_learned" --json
memesh recall --tag "type:session_keypoint" --json
If MCP user_patterns tool is available, also run it for work pattern analysis:
user_patterns: {}
Step 2: Analyze and report
From the recalled data, compute and present:
## Memory Health Report
### Overview
- Total entities: N
- Last 30 days active: N (N%)
- Knowledge types: N decisions, N patterns, N lessons, N auto-tracked
### Health Score: N/100
- Activity: N% (accessed in last 30 days)
- Quality: N% (high confidence, well-tagged)
- Freshness: N% (new this week)
- Self-Improvement: N% (lessons learned ratio)
### Quality Issues Found
**Stale (not accessed 30+ days, low confidence)**
- "entity-name" — confidence: N% — Suggest: archive?
**Verbose (5+ observations)**
- "entity-name" (N observations) — note it; there is no one-entity compression
command any more. If the noise is spread across many episodic entries,
suggest `memesh dream run` instead.
**Potential conflicts**
- "entity-A" vs "entity-B" — contradicting decisions
**Noise ratio**
- N% auto-tracked (session_keypoint, commit) vs N% intentional knowledge
- If noise > 80%: recommend more deliberate `memesh remember` usage
### Recommended Actions
1. `memesh forget --name "old-design"` (superseded)
2. `memesh dream run --project myapp` (propose digests for the noisy clusters, then review)
3. `memesh remember ...` (knowledge gap in [area])
Step 3: Execute approved actions
Present the report first. Ask which actions to execute. Then run the commands:
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 124 lines · 64 tokens per session scan A 64231229f814
memesh-review is a skill published in the GitHub repository PCIRCLE-AI/memesh (15 stars, last pushed yesterday), licensed MIT. It adds 64 tokens to every session and 933 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
kip-cognitive-nexus
Persistent graph-based memory for AI agents via KIP (Knowledge Interaction Protocol). Provides retrieval-first memory operations (KQL), durable writes (KML), schema discovery (META), and memory hygiene patterns. Use whenever the agent needs to consult or update persistent memory, especially for: remembering user…
forget
Delete a Wenlan memory by exact id from Codex. Destructive and confirmation gated. Invoked as /forget .
handoff
End a Codex work session. Stores durable captures, writes a narrative session log, and automatically applies typed item-level deltas to the current Space Brief. Invoked as /handoff.
capture
Save a durable memory to Wenlan from Codex. Use proactively when the user states a preference, makes a decision, corrects you, or shares a durable fact. Invoked as /capture .
recall
Search Wenlan's local memory from Codex by query. Targeted lookup, not session orientation. Invoked as /recall .
lint
Lint Wenlan memory quality and hygiene, check vocabulary, or resolve every finding into a ready repair, review item, system action, or blocker.