fix-github-pr
Pass
Audited by Gen Agent Trust Hub on Feb 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by ingesting untrusted data from external sources.
- Ingestion points: Fetches PR comments via
gh apiand CI logs viagh run view. - Boundary markers: Lacks explicit delimiters or instructions to ignore potential commands within the fetched content.
- Capability inventory: Includes file system modification, git commits/pushes, and shell command execution (
bundle exec rspec). - Sanitization: No sanitization of external text is performed before the agent evaluates and implements requested changes.
- [COMMAND_EXECUTION]: The skill performs several shell operations using
git,gh, andbundle exec. This includes creating worktrees, committing code, and executing the Ruby test suite (rspec), which may run code provided within the Pull Request.
Audit Metadata