Skills
skills/pitzcarraldo/skills/review-and-fix/Gen Agent Trust Hub

review-and-fix

Pass

Audited by Gen Agent Trust Hub on Apr 16, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data from the CodeRabbit CLI output and instructions from a local AGENTS.md file. This represents an attack surface where untrusted data could attempt to influence agent behavior.\n
  • Ingestion points: Processes output from coderabbit review --agent and reads instructions from AGENTS.md (Step 0).\n
  • Boundary markers: None explicitly defined for tool output parsing; the skill relies on the structured nature of the --agent flag.\n
  • Capability inventory: Includes shell execution (CodeRabbit, pnpm, tsc) and file modification via the Edit tool.\n
  • Sanitization: The skill mandates a robust validation step (Step 3.2) requiring the agent to manually verify every finding against the actual source code and context before applying any fix.\n- [COMMAND_EXECUTION]: The skill uses several shell commands to interact with the environment and project tooling.\n
  • Evidence: Executes coderabbit review, coderabbit update, and project validation commands such as pnpm lint, pnpm build, and pnpm test.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 16, 2026, 06:57 AM