github-pr-comment-analyzer
Pass
Audited by Gen Agent Trust Hub on Mar 9, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from GitHub PR comments and incorporates it into analysis reports.
- Ingestion points: Comment bodies are fetched via GraphQL queries using the GitHub CLI in 'SKILL.md' (Step 1.2) and 'references/github_api_reference.md'.
- Boundary markers: The skill uses markdown blockquotes ('>') to delimit comment content in the generated report (Step 4.2), which provides minimal isolation for the untrusted content.
- Capability inventory: The skill executes local shell commands using the 'gh' CLI and performs file write operations to save reports.
- Sanitization: There is no explicit sanitization or escaping of the ingested comment text before it is processed or written to the final markdown report.
- [COMMAND_EXECUTION]: The skill relies on the GitHub CLI ('gh') to interact with the GitHub API.
- Evidence: Multiple instances of 'gh pr view', 'gh api graphql', and 'gh pr checks' are used across 'SKILL.md' and the reference documentation.
- Context: These commands are used for their intended purpose of gathering pull request metadata and comments for analysis.
Audit Metadata