The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's bootstrap and health check processes use pip to install the tree-sitter, tree-sitter-typescript, and tree-sitter-javascript packages from the well-known PyPI registry. These libraries are used for the skill's primary purpose of performing Abstract Syntax Tree (AST) analysis on source code.
[COMMAND_EXECUTION]: The skill utilizes subprocess.run and os.execv for several infrastructure and functional tasks:
bootstrap.py uses os.execv to re-execute itself if a newer Python interpreter (3.10+) is found on the system path, ensuring compatibility with its dependencies.
session-health-check.py and other hooks execute git commands to detect code changes and pip commands to maintain the skill's execution environment.
judge-comments.py invokes the claude CLI in a restricted mode (no tools, no MCP, single turn) to perform out-of-band analysis of source code comments.
[PRIVILEGE_ESCALATION]: The script _bootstrap/permissions.py modifies the agent's platform configuration (settings.json or settings.local.json) to add allow rules for the skill's own Python scripts (bootstrap.py and review-files.py). This is an intentional design choice to prevent the agent from prompting the user for permission every time the skill's core scripts are executed.
[DYNAMIC_EXECUTION]: The skill uses __import__ in _bootstrap/dependencies.py to dynamically check for the presence of required packages before attempting to use them for AST-based code audits.

coding-standards