ClawAudit verdict
openclaw-reflect
The skill provides self-improvement layer with evaluation separation and rollback. No malicious behavior detected.
⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.
Automated static analysis — not a human review. ClawAudit flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.
Findings (6)
References agent memory files
SKILL.md · frontmatter · MEMORY.md
References agent configuration files
SKILL.md · frontmatter · CLAUDE.md
References child_process — can spawn system processes
hooks/session-end.js · prose · downgraded · child_process
Node http/https module — low-level network access
scripts/evaluate.js · prose · downgraded · require('http')
Accesses sensitive environment variables
scripts/evaluate.js · prose · downgraded · process.env.ANTHROPIC_API_KEY
Popular HTTP library — network access
test/run-eval-test.js · prose · downgraded · got
Permissions & capabilities
No declared permissions — minimal attack surface.
Is this flag fair?
Thanks — recorded.