skills/mattstruble/skills/git-pr/Gen Agent Trust Hub

git-pr

Pass

Audited by Gen Agent Trust Hub on Jun 28, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements a robust security model by requiring explicit human confirmation ('STOP and wait') in Step 4 before executing commands that affect the remote repository in Step 5. This prevents unauthorized code pushes or PR creations.\n- [INDIRECT_PROMPT_INJECTION]: The skill identifies and ingests data from external sources that may be attacker-controlled.\n
  • Ingestion points: The workflow reads repository history ('git log'), diffs ('git diff'), remote PR metadata ('gh pr list'), and local template files ('.github/PULL_REQUEST_TEMPLATE.md').\n
  • Boundary markers: There are no instructions provided to separate the untrusted data from the prompt logic using delimiters or specific warnings.\n
  • Capability inventory: The skill uses the 'git' and 'gh' CLI tools to perform write operations ('git push', 'gh pr create') to the remote server.\n
  • Sanitization: No evidence of sanitization, escaping, or validation of commit messages or template contents is found in the instructions.\n
  • Note: The severity is mitigated to safe levels because the agent is instructed to stop and wait for user confirmation of the generated content before proceeding.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 28, 2026, 08:28 PM
Security Audit — agent-trust-hub — git-pr