code-review-playbook

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_CONTEXT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from GitHub pull requests, including titles, descriptions, comments, and code diffs. This creates a surface where an attacker could embed malicious instructions in a PR to influence the agent's behavior during review. Ingestion points occur in 'scripts/review-pr.md' and 'scripts/run-pr-checks.py' via 'gh' commands. No boundary markers or sanitization are present to mitigate this risk. The agent's access to 'WebFetch' and 'Bash' tools increases the potential impact of such an injection.
  • [DYNAMIC_CONTEXT_INJECTION]: The 'scripts/review-pr.md' file employs the '!' syntax to execute shell commands at load time, specifically 'gh pr list' to fetch recent pull requests. This provides automated context retrieval from the local environment.
  • [COMMAND_EXECUTION]: The skill utilizes shell commands to execute developer tools such as the GitHub CLI ('gh'), 'ruff', 'eslint', and 'biome' for code analysis and metadata retrieval.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 02:16 AM
Security Audit — agent-trust-hub — code-review-playbook