ClawAudit verdict
claude-code-skill
openclaw-claude-code
Skill controlling Claude Code via MCP from within OpenClaw agents; all operations are local Claude Code sessions with user-controlled budgets and permission modes, matching the stated agent-integration purpose.
⚠ 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)
Attempts to modify skills or system prompts
SKILL.md · code · Replace system prompt
HTTP request to bare IP address — common in malicious payloads
SKILL.md · code · http://127.0.0.1
References sudo — requests elevated privileges
SKILL.md · code · sudo
Pipe to bash — executes piped content as shell commands
SKILL.md · prose · downgraded · | Bash
Popular HTTP library — network access
package.json · prose · downgraded · node-fetch
Why the tier is capped
Execution sink present in raw bytes (Hard Floor: class A/B). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.
Permissions & capabilities
Requires 1 system binary.
network_in Is this flag fair?
Thanks — recorded.