Memory Hygiene & Truth Ledger 20260717 — Release Notes v1.0
Release Date: 2026-07-17
Price: $4.99
Category: memory
Status: DOCS_COMPLETE — awaiting Rooster listing
What It Does
Memory Hygiene & Truth Ledger converts agent memory governance notes into a deterministic PASS, REVIEW, or BLOCKED hygiene report. It checks for source references, record ownership, verification dates, freshness windows, truth status, contradiction checks, superseded markers, retention rules, and review cadence — flagging risks like stale records, contradictory facts, missing owners, silent overwrites, and unverified memory. Running fully offline, it never reads or modifies your actual memory files.
Use Cases
- Agent operators auditing their memory governance notes before a scheduled memory review, getting a scored evidence report that shows which hygiene signals are covered and which gaps need attention.
- Teams establishing contradiction checks and retention rules before handing off agent memory to a new owner, ensuring there is documented evidence of human approval and ownership records.
- Compliance-minded operators who need a repeatable truth-ledger audit trail showing memory health across time, covering freshness, contradiction, and retention evidence without exposing actual memory contents.
Requirements
- Python: 3.8 or later (included with OpenClaw)
- Dependencies: None — Python standard library only (
argparse,json,re,pathlib) - API keys: Not required
- OpenClaw version: 1.0+
- Network access: Not required — runs entirely offline
Example Usage
# Analyse a local notes file
python3 scripts/memory_hygiene_truth_ledger.py --notes memory-review-notes.txt --json
# Pass inline text directly
python3 scripts/memory_hygiene_truth_ledger.py --text "source reference, record owner, record date..." --json
# Run built-in self-test
python3 scripts/memory_hygiene_truth_ledger.py --self-test --json
Expected Output (PASS)
{
"findings": [],
"missing_signals": [],
"mutated_files": [],
"ok": true,
"recommendations": [
"Keep this result advisory pending independent Viper review."
],
"risks_found": [],
"score": 108,
"signals_found": [
"source reference", "record owner", "record date",
"last verified", "freshness window", "truth status",
"contradiction check", "superseded marker", "retention rule",
"review cadence", "human approval", "change evidence"
],
"status": "PASS"
}
Expected Output (REVIEW — partial signals)
{
"mutated_files": [],
"ok": true,
"risks_found": ["stale record"],
"score": 38,
"signals_found": ["source reference", "record owner", "record date"],
"missing_signals": ["last verified", "freshness window"],
"status": "REVIEW"
}