Works with the AI Agents you already use
Memory Health Pro — Audit Your Agent Persistent Memory
Audit AI agent memory stores for leaked secrets, contradictory rules, stale file paths, and metadata bloat.
$13
Memory Health Pro — Audit Your Agent Persistent Memory
Example session with this skill installed
Run a memory health audit on the .agent/memory directory. Check for any leaked keys, stale file paths from my old laptop, and redundant rules.
- Read your context and instructions
- Compiled the memory health pro
I have completed the health audit of the .agent/memory directory. I scanned 4 files, verified 12 system paths, and checked for credential patterns. I found one active contradiction regarding file writing permissions and two stale paths referencing a Documents/old_project directory that no longer exists.
Memory audit: .agent/memory (2025-05-22)
| File | Size | PII | Bloat | Duplicates | Dead refs | Verdict | Action |
|---|---|---|---|---|---|---|---|
| core_rules.md | 2.4 KB | none | ok | 1 conflict | none | FAIL | Resolve conflict: "always ask" vs "silent write" |
| paths.md | 0.8 KB | none | ok | none | 2 stale | REVISE | Delete refs to /Users/dev/old_project/ |
| secrets.env | 0.2 KB | sk-... | ok | none | none | FAIL | CRITICAL: Delete file; rotate OpenAI key found at line 2 |
| scratchpad.log | 15 KB | none | high | none | none | REVISE | Distill into 3 rules and delete raw logs |
Summary: 0 PASS, 3 REVISE, 1 FAIL — deletions proposed:
- Credential at
secrets.env:2(Leaked) - Stale paths in
paths.md(Invalid local references) - Conflicting write instruction in
core_rules.md - 80% of
scratchpad.log(Redundant session context)
Next steps
- Rotate your OpenAI key immediately as it is stored in plaintext in the memory store.
- Delete the
secrets.envfile and move the key to a secure environment variable. - Run
rm .agent/memory/scratchpad.logafter I distill the relevant logic intocore_rules.md.
Connects securely to your tools. The creator never sees your data.
About this skill
Persistent memory stores in AI agents often become dumping grounds for sensitive data, stale paths, and conflicting instructions. This skill audits memory directories to identify security risks and operational bloat that degrades performance. It systematically evaluates memory files to ensure they remain lean, compliant, and accurate across sessions. ## What it does - PII & credential scanning identifies leaked API keys, tokens, emails, and phone numbers that create security vulnerabilities. - Bloat detection flags oversized files and session logs that should have been distilled rather than archived. - Conflict analysis detects contradictory rules and duplicate instructions that lead to unpredictable agent behavior. - Dead reference checking verifies that every file path, command, and tool mentioned in memory actually exists on the host system. - Retention auditing ensures every memory file has a clear purpose and hasn't outlived its usefulness. ## How it works 1. Inventory: The agent lists all memory files, sizes, and timestamps to establish an audit baseline. 2. Scan: It performs deep searches for pattern-based credentials, PII, and technical contradictions. 3. Verification: The agent runs read-only checks on all referenced system paths and tool versions. 4. Reporting: You receive a structured audit table with PASS, REVISE, or FAIL verdicts for every file. ## Frameworks & tools This skill is platform-agnostic and works with any agent harness that uses local or remote file-based memory stores (Markdown, JSON, or text). ## Why this beats prompting it yourself Standard prompts often miss subtle contradictions across multiple files or fail to verify if referenced paths actually exist. This skill enforces a strict verification checklist and prevents the "secret leak" problem where an auditor accidentally repeats a found API key in its own report. ## Use cases - Auditing a memory folder before committing it to a shared repository. - Cleaning up an agent that has become slow or confused due to context window saturation. - Ensuring GDPR compliance by identifying unneeded personal data in long-term stores. - Validating agent configurations after migrating to a new machine or environment. ## Known limitations The skill requires read access to the memory directory and the ability to execute basic file existence checks via the terminal. It cannot rotate leaked keys for you. ## Upgrade path This skill is the lightweight audit companion of Obsidian Hippocampus MCP — Persistent Agent Memory ($60), the full memory engine: semantic recall, Ebbinghaus forgetting, bidirectional Obsidian vault sync, and a bundled CLI + native MCP server mode. View it here: https://www.agensi.io/skills/obsidian-hippocampus-mcp-persistent-agent-memory
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.
Requires read access to memory files and terminal access for path/tool verification. Does not automate the rotation of compromised credentials.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 20 days ago