memory-dream-governance

Memory governance skill for OpenClaw with a dry-run → audit gate → guarded apply pipeline.

What it does

• Generates a governance plan from memory markdown scope (dry_run.py)
• Audits risk with policy thresholds (audit.py)
• Auto-applies only when decision is GO (apply.py)
• Writes rollback snapshots before mutation (snapshot.py)
• Supports manual rollback (rollback.py)
• Includes a cron wrapper for unattended runs (wrappers/cron-memory-dream-governance.sh)

Safety model

• Writable scope is policy-whitelisted (MEMORY.md, memory/**/*.md, index files)
• Forbidden files are blocked by policy (SOUL.md, AGENTS.md, TOOLS.md, openclaw.json)
• REVIEW/BLOCK never applies write actions
• Snapshot metadata is persisted in apply reports for rollback traceability

Repository layout

.
├── SKILL.md
├── config/
├── scripts/
├── references/
└── wrappers/

Quick start

# 1) Dry-run plan
python3 scripts/dry_run.py --workspace /root/.openclaw/workspace --skill-root .

# 2) Audit decision
python3 scripts/audit.py --plan <plan.json> --skill-root .

# 3) Apply (GO only)
python3 scripts/apply.py --plan <plan.json> --audit <audit.json> --workspace /root/.openclaw/workspace --skill-root . --apply

Cron wrapper behavior

wrappers/cron-memory-dream-governance.sh executes:

1. dry-run plan generation
2. audit scoring and decision
3. auto-apply when decision=GO and policy/environment allows
4. escalation report generation when decision is not GO

Environment override:

MEMORY_DREAM_APPLY=1 bash wrappers/cron-memory-dream-governance.sh

Topics

Recommended GitHub topics:

• openclaw
• openclaw-skill
• memory-governance
• audit-gate
• cron-automation
• python

Release

Initial stable release: v1.0.0

• Auto-apply on GO
• Escalation on REVIEW/BLOCK
• Snapshot-in-report traceability