skills/vdustr/skills/vp-git/Gen Agent Trust Hub

vp-git

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data from Git repositories, including commit history, pull request discussions, and repository metadata, which could theoretically contain malicious instructions.
  • Ingestion points: Processes local history, remote state, and hosting-platform data as separate sources of truth (SKILL.md, references/pull-requests.md).
  • Boundary markers: Explicitly mandates "read-only inspection" and "explicit authorization" for any writes to history, remotes, or PR states (SKILL.md).
  • Capability inventory: Performs git commits, branch cleanup, remote synchronization, and pull request lifecycle management (references/cleanup.md, references/pull-requests.md).
  • Sanitization: Requires verification of exact diffs and inspection of repository-specific instructions and templates before making changes (references/commits.md, references/repositories.md).
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch .gitignore templates from official sources (e.g., GitHub's template repository) to ensure ecosystem defaults are maintained.
  • [SAFE]: The instructions reinforce security best practices, such as warning against using ignore rules to hide already-tracked secrets and advising against installing unknown tools to bypass commit hooks.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 09:02 AM
Security Audit — agent-trust-hub — vp-git