Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Skill Safety Auditor
v1.0.0Perform comprehensive security audits on skills to identify vulnerabilities, unsafe patterns, and compliance issues. Use when auditing skills for security, c...
⭐ 0· 190·0 current·0 all-time
byhaidong@harrylabsj
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name/description match the observed assets: the package is an instruction-only skill implemented as shell scripts that perform static checks (grep/find/permission checks/dependency lookups). No unrelated credentials, binaries, or network endpoints are requested.
Instruction Scope
SKILL.md directs the agent to run scripts against a provided skill directory and produce console/JSON reports. The scripts read the entire target directory, search files for patterns (secrets, eval, exec, http://, etc.), and write JSON reports into a local data directory. This is expected for an auditor, but it means the skill will read arbitrary files under the supplied path and write report files — review what path you point it at.
Install Mechanism
No install spec; skill is instruction+scripts only. That is low risk compared to arbitrary downloads, but running provided shell scripts executes code on your system — treat as code you must review/inspect before executing.
Credentials
The skill declares no required env vars, which is appropriate, but scripts optionally use AUDIT_DATA_DIR and list-audits.sh invokes jq without declaring jq as a required binary. Also the included data/AUDIT-20260313-001.json reports critical 'secrets' findings inside the skill's own scripts (SEC-001, SEC-002) — having hardcoded secrets inside an auditor package is unexpected and a red flag.
Persistence & Privilege
No 'always: true' privilege; the skill does not attempt to modify other skills or system-wide settings. test.sh may change executable bits for the packaged scripts (chmod +x), and audit scripts write report files to their own data directory — these are local effects but should be noted.
Scan Findings in Context
[SEC-001] unexpected: The included audit report (data/AUDIT-20260313-001.json) flags a critical 'secrets' finding in scripts/quick-scan.sh. An auditing tool packaging hardcoded secrets in its own scripts is unexpected and should be investigated or removed before running.
[SEC-002] unexpected: The same audit report flags a critical 'secrets' finding in scripts/audit-skill.sh. The presence of these findings inside the skill package itself is a security concern and reduces trust in the distributed artifact.
What to consider before installing
This skill performs static audits and that matches its description, but treat the packaged scripts as code you must inspect before executing. Recommended steps before installing or running: 1) Open scripts/audit-skill.sh, quick-scan.sh, and test.sh and search for any hardcoded credentials or outbound network calls; remove any secrets found. 2) Because list-audits.sh uses jq but the skill doesn't declare jq as required, ensure jq is available or modify the script to handle its absence. 3) Run the scripts in an isolated/sandbox environment (or container) pointed at a non-sensitive test directory first — the scanner reads arbitrary files and will write JSON reports into the skill's data directory. 4) Consider removing or cleaning the included data/AUDIT-*.json files (they appear to contain findings that reference secrets in the package). 5) If you plan to allow autonomous agent invocation, only do so after cleaning the package; otherwise restrict invocation until you vet the code. If you want, I can highlight the exact lines in the scripts that look suspicious or walk through how to run them safely in a container.scripts/audit-skill.sh:169
Dynamic code execution detected.
scripts/quick-scan.sh:46
Dynamic code execution detected.
Patterns worth reviewing
These patterns may indicate risky behavior. Check the VirusTotal and OpenClaw results above for context-aware analysis before installing.Like a lobster shell, security has layers — review code before you run it.
latestvk97dzssffp2zxfzazmg5s91s6x82v45p
License
MIT-0
Free to use, modify, and redistribute. No attribution required.