fix-dep-vulns
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill requires the agent to read and interpret untrusted external data from security advisories (e.g., GHSA, CVE, OSV) to determine exploitability and decide on dismissals.
- Ingestion points: Security advisory descriptions and metadata are read in 'SKILL.md' (Step 1 and Step 3d) to guide the triage process.
- Boundary markers: There are no instructions to use delimiters or ignore instructions within the external advisory text, creating a surface for potential instruction injection.
- Capability inventory: The skill can modify project lockfiles and configuration files (package.json, pyproject.toml, pnpm-workspace.yaml), execute package manager commands (update, add, remove), and run shell commands (curl, grep, jq).
- Sanitization: No sanitization is performed on the text of external advisories before the agent processes them for decision-making.
- [EXTERNAL_DOWNLOADS]: The skill uses
curlto fetch package metadata from the official PyPI registry API ('pypi.org') as described in 'references/python.md'. This is a standard operation for dependency management and uses a well-known, trusted service.
Audit Metadata