dependency-vulnerability-triage

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill implements standard security practices for managing dependency vulnerabilities and no malicious intent or suspicious code patterns were detected.
  • [COMMAND_EXECUTION]: The skill invokes common development tools such as npm audit, npm test, and git to perform its tasks. These operations are conducted locally and are necessary for the tool's stated purpose of vulnerability management. Evidence: execSync usage in scripts/parse-audit.ts and shell commands in scripts/apply-patches.sh.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources (npm audit reports) to generate patch plans. While this represents an ingestion of untrusted data, the risk is inherent to the intended use case and the skill uses standard parsing methods. Ingestion points: scripts/parse-audit.ts. Capability inventory: npm audit fix, git commit. Sanitization: Standard JSON parsing and jq utility usage.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 06:14 PM
Security Audit — agent-trust-hub — dependency-vulnerability-triage