The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill generates Python scripts at .claude/hooks/ and executes them using uv run as part of its mandatory verification process. The generated code templates also utilize subprocess.run with shell=True to execute project-level commands.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input from the $ARGUMENTS variable to determine the logic and parameters of generated hooks.
Ingestion points: User-provided behavior descriptions enter the context via $ARGUMENTS in SKILL.md.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands within the user input.
Capability inventory: The skill has the ability to perform network requests (WebFetch), write files to the project directory, and execute shell commands (uv run).
Sanitization: There is no evidence of input validation or escaping before the user-supplied text is used to generate executable scripts.
[EXTERNAL_DOWNLOADS]: The skill retrieves lifecycle event specifications and protocol documentation from the official Claude Code website.

hooks-create