The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/pre-review.py utilizes subprocess.run to execute the ruff linter for automated code analysis. This is a functional requirement for the skill. The execution is performed by passing arguments as a list and does not involve a shell, effectively mitigating potential command injection risks.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes user-provided source code, which serves as an ingestion point for untrusted data.
- Ingestion points: The scripts/pre-review.py script reads user-provided files, and the contents are subsequently processed by the AI agent for review.
- Boundary markers: The instructions in SKILL.md use structured headers and a clear multi-step review process to maintain focus.
- Capability inventory: The skill uses Python's ast module for structural analysis and the ruff binary for linting; it does not have file-write or network capabilities.
- Sanitization: No explicit sanitization or escaping of the user-provided code is performed before it is interpolated into the agent's context. While this represents a standard attack surface for AI tools, there are no specific malicious instructions identified in the skill itself.

clean-code-reviewer