detecting-security-vulnerabilities
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- EXTERNAL_DOWNLOADS (LOW): The skill uses
npxto fetch and execute remote packages includingsnyk,eslint, and@semgrep/semgrep. While these are industry-standard security tools, the use ofnpxinvolves downloading and executing code from the npm registry. This is downgraded to LOW severity as it is the primary intended function of the skill.\n- COMMAND_EXECUTION (LOW): The skill instructs the agent to run various shell commands (git diff,find,grep,gitleaks) to audit the filesystem for vulnerabilities. These operations are essential for a security scanner and are considered acceptable within this context.\n- INDIRECT_PROMPT_INJECTION (LOW): The skill has a surface for indirect prompt injection as it ingests untrusted code and scanner results to generate a report.\n - Ingestion points:
SKILL.md(viagit diff,find,grep, and JSON scanner outputs).\n - Boundary markers: Absent; the skill does not use specific delimiters to isolate untrusted data from its reporting logic.\n
- Capability inventory: Shell execution (
git,grep,npx,find), filesystem access, and network access (vianpm/npx).\n - Sanitization: Absent; findings are directly interpolated into the generated security report without validation.
Audit Metadata