The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run in several scripts (detect_hook_tool.py, lint_orchestrator.py) to execute formatting and linting tools. These implementations are secure as they either validate the tool name against an allowed set or use a hardcoded whitelist (TOOL_COMMANDS) to prevent arbitrary command injection from untrusted project configurations.
[SAFE]: Configuration parsing in discover_linters.py uses ruamel.yaml with typ='safe', which effectively mitigates the risk of arbitrary code execution during the deserialization of project-level .pre-commit-config.yaml files.
[SAFE]: The skill includes an extensive referential knowledge base for common Python security (Bandit) and type-checking (MyPy, Ruff) rules. This content is strictly informative and serves to assist the AI agent in resolving legitimate code quality issues.
[SAFE]: The agent installation script (install_agents.py) is transparent and includes integrity checks using SHA256 hashing to ensure that agent files are not modified or overwritten without verification.

holistic-linting