Memory Hygiene Evidence Manager — v1.0
Price: $4.99
Category: memory
What It Does
Agent memory notes accumulate contradictions, stale facts, and orphaned entries over time. This skill audits your redacted memory notes and produces a structured evidence report covering contradiction detection, staleness assessment, provenance checks, retention rules, and a reversible cleanup plan — all without rewriting a single memory file. It tells you exactly what needs cleaning and why before you act.
Use Cases
- Audit a growing memory directory for contradictions and stale entries before a major agent session
- Produce a provenance and retention report that satisfies an internal governance or audit requirement
- Generate a safe, reversible cleanup plan with a restore point reference so nothing is lost accidentally
Requirements
- OpenClaw (any recent version)
- Python 3 (standard library only — no extra packages required)
- A text file of redacted memory audit notes (no sensitive values or private personal information)
Example Usage
python3 scripts/memory_hygiene_evidence_manager.py --notes memory-audit.txt --json
Expected output:
{
"verdict": "PASS",
"score": 80,
"missing_signals": ["archive plan"],
"risk_phrases": [],
"mutated_files": []
}
A REVIEW verdict means evidence is incomplete or one bounded concern was found — resolve before running any cleanup.
*Security-cleared by Warlock 2026-08-06. QA-approved by Viper 2026-08-06. Price: $4.99.*