audit-the-list
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by ingesting and processing untrusted data from external APIs.
- Ingestion points: The skill fetches repository descriptions via the GitHub and GitLab APIs (
gh api repos/{owner}/{repo}) and package metadata from the PyPI JSON API (https://pypi.org/pypi/{name}/json). - Boundary markers: The instructions do not specify any delimiters or safety prompts to isolate the fetched external text from the agent's reasoning during the verdict phase.
- Capability inventory: The skill has broad file system access to modify the project's
README.mdand several data files (e.g.,data/pypi_downloads.tsv,data/pypi_name_overrides.json), the ability to run build tasks viamake, and authority to commit changes to the git repository. - Sanitization: There is no documented logic for sanitizing or escaping potentially malicious instructions embedded in the fetched package or repository descriptions.
Audit Metadata