git-pr
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, specifically pull request and merge request comments and review bodies, which could contain malicious instructions intended to influence the agent's actions during the PR lifecycle. This risk is a necessary part of the skill's primary function. 1. Ingestion points: Data is fetched via gh pr view, gh api graphql, and glab mr view as seen in SKILL.md and references/pr-comment-workflow.md. 2. Boundary markers: The skill does not explicitly utilize boundary markers or delimiters for external comment data. 3. Capability inventory: The skill has extensive write capabilities, including merging PRs, posting comments, and resolving threads as detailed in SKILL.md and references/line-comments.md. 4. Sanitization: There is no explicit sanitization logic for external text content in the provided scripts.
- [COMMAND_EXECUTION]: The skill generates and facilitates the execution of complex shell commands via the gh and glab CLIs. These include subshells for dynamic argument resolution, piping to jq for data filtering, and GraphQL mutations for thread management.
- [EXTERNAL_DOWNLOADS]: Provides instructions for installing the GitHub Copilot CLI tool via the npm package manager.
Audit Metadata