Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed remote LLM-based pre-scan tool for SKILL.md files, with no evidence of hidden persistence, destructive behavior, or privilege escalation.
Install only if you are comfortable sending the target SKILL.md, and a manifest of nearby file names and sizes, to the configured LLM provider. Do not scan secrets or proprietary material unless that provider is approved for it; use the confirmation prompt carefully and reserve --yes for controlled CI use.
The description says the scan is performed 'locally,' but the document itself states that the full SKILL.md is sent to a remote third-party LLM endpoint. That mismatch can mislead users into disclosing proprietary or sensitive content under the false impression that analysis stays on-device, and the claimed equivalence to the real scanner may also overstate security coverage.
Referenced artifact was not completely inspected
This tool reads the target `SKILL.md` and **sends its full contents to the
Referenced artifact was not completely inspected
This tool reads the target `SKILL.md` and **sends its full contents to the
The skill declares no explicit tool scope or permissions even though its documented behavior requires reading local files, accessing environment variables for API keys, and making outbound network requests. This is risky because users and automated reviewers cannot quickly verify the intended execution boundary, increasing the chance of unintended data exposure or over-broad execution in agent environments.
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.
if not args.yes:
if not sys.stdin.isatty():
print(
"\nError: refusing to transmit without confirmation in a "
"non-interactive session. Re-run with --yes to proceed.",
file=sys.stderr,
)
Skill requests more permissions than appear necessary for its stated functionality. Review if elevated access is justified.
- **instruction_scope** — Are the instructions bounded and user-controlled?
- **install_mechanism** — Is the install path transparent and verifiable?
- **environment_proportionality** — Are credentials/environment requirements proportionate?
- **persistence_privilege** — Does the skill persist or escalate privileges?
### Key Rule from ClawHub's Scanner
No suspicious patterns detected.