ClawAudit verdict
architecture-critic
Adversarial pre-build review skill that calls Claude via ANTHROPIC_API_KEY to review proposals before code is written; credential use is transparent and purpose-aligned, no exfiltration or manipulation.
⚠ 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 (5)
Accesses OpenClaw config/secrets directly
clawhub.yaml · prose · downgraded · ~/.openclaw/openclaw.json
Instructs covert action — may act without user awareness
references/payment-flows.md · prose · downgraded · silently
subprocess execution — runs system commands from Python
scripts/run-critic.sh · prose · downgraded · subprocess.run(
Opens WebSocket connection
references/ai-builds.md · prose · downgraded · WebSocket
Python os.environ.get — reads environment variable
scripts/run-critic.sh · prose · downgraded · os.environ.get(
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class D). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
No declared permissions — minimal attack surface.
Is this flag fair?
Thanks — recorded.