Security audit
Security checks for vulnerabilities and agentic risk
ClawScan marks this skill suspicious because it reports device-pairing approvals that can change authorization state, including a command described as preview-only; the scan is partial.
Consider the reported pairing-approval behavior carefully before installing. In particular, the scan says the `--latest` command is not read-only despite being described as a preview. The verdict is suspicious, not malicious. Coverage is incomplete because a referenced local path was missing, so the scan does not establish that all related material was inspected.
Device listing and node status checks support the declared diagnostic purpose. However, openclaw devices approve <requestId> performs a state-changing approval not disclosed by that purpose. The comment describing approve --latest as preview-only does not establish that behavior, and the explicit request-ID approval command independently demonstrates the mismatch.
The skill incorrectly presents devices approve --latest as a read-only preview, although it approves the latest pending request. An agent following this sequence could authorize an unintended or attacker-controlled device before performing the exact-request approval; subsequent identity checks cannot undo that authorization.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
openclaw devices list
openclaw devices approve --latest # preview only; exits without approval
openclaw devices approve <requestId>
openclaw nodes status