Use this checklist if you have already applied (or are currently applying) the SlowMist OpenClaw Security Practice Guide v2.7 and want to add the AI SAFE² Framework’s external enforcement layer.
If you are starting from scratch, complete the SlowMist guide first, then return here.
scanner.py against your OpenClaw data directory — record the baseline risk scoreopenclaw_memory.md from examples/openclaw/ in the AI SAFE² repo~/.openclaw/workspace/openclaw_memory.md. Confirm when loaded.”gateway/README.md for configuration optionssudo command) — confirm gateway logs the request and routes for human approvalscanner.py execution to SlowMist’s nightly audit cron (or as a separate parallel cron)/tmp/openclaw/security-reports/scanner-$(date +%Y-%m-%d).txtred-team-schedule.md)| Task | Frequency | Owner |
|---|---|---|
| Review nightly audit + scanner report | Daily | Operator |
| Review Gateway audit logs for anomalies | Weekly | Operator |
| Check skill baseline against approved inventory | Weekly | Operator |
| Run Scanner manually after any skill install | Per event | Operator |
| Red-team exercise (SlowMist curriculum) | Quarterly | Security team |
| Memory Vaccine review + update | On SlowMist guide update | Security team |
| Gateway config review | On OpenClaw major release | Security team |
| Deployment registry audit | Semi-annual | Security lead |
| Threat model review | Annual | Security lead |
*Cross-reference: README.md |
safe2-for-slowmist-overlay.md* |