ClawAudit verdict
dual-brain
The skill runs a background daemon that intercepts every user message from OpenClaw session files and forwards it to external LLM providers (Moonshot/Kimi, OpenAI, Groq, Ollama) without per-message user confirmation, constituting ongoing conversation exfiltration to third parties beyond the user's control.
⚠ 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.
Permission integrity
package_install
Findings (6)
Recursive delete from root or home — destructive command
QUICKSTART.md · code · rm -rf ~
References sudo — requests elevated privileges
QUICKSTART.md · code · sudo
Accesses system credential store
BUILD-SUMMARY.md · prose · downgraded · keychain
References child_process — can spawn system processes
src/cli.js · prose · downgraded · child_process
References agent memory files
BUILD-SUMMARY.md · prose · downgraded · MEMORY.md
Node http/https module — low-level network access
src/daemon.js · prose · downgraded · require('http')
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.
package_install Is this flag fair?
Thanks — recorded.