The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run and os.execv for environment management, such as switching to compatible Python versions and executing its own internal linting scripts. These actions are consistent with its stated purpose of managing a complex linting infrastructure.
[EXTERNAL_DOWNLOADS]: During the bootstrap process, the skill downloads and installs well-known code parsing libraries (tree-sitter and associated grammars) from the official Python Package Index (PyPI). This is required for its AST-based linting rules to function correctly.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection (Category 8) due to its core functionality of reading and analyzing untrusted source code from the project tree.
Ingestion points: Code is ingested by hooks/review-files.py and the orchestrator logic for review and fixing tasks.
Boundary markers: The skill implements a multi-agent pipeline where specialist workers are constrained by specific JSON-only output requirements and role-based instructions.
Capability inventory: The orchestrator agent is capable of dispatching sub-agents and applying modifications to the filesystem via the agent's native tools.
Sanitization: The skill processes code primarily as text for transformation by specialist LLM workers, using multi-step verification passes to ensure compliance.

coding-standards