The Agent Skills Directory

[COMMAND_EXECUTION] (MEDIUM): The skill's primary function is to generate and run Python scripts locally using uv run. This represents a significant capability for arbitrary code execution, though it is the stated purpose of the skill.
Evidence: Workflow step 5 in SKILL.md and the provided script template in references/script-template.md demonstrate the intent to execute generated logic.
[REMOTE_CODE_EXECUTION] (LOW): The skill uses the uv tool to automatically download and install Python packages listed in the script's metadata. While the template uses standard packages, this mechanism allows the execution of code from external registries.
Evidence: Dependency list in references/script-template.md includes click, rich, and requests which are fetched at runtime.
[EXTERNAL_DOWNLOADS] (LOW): The skill performs external network requests to fetch dependencies and linting tools from PyPI during the development and execution phases.
Evidence: Workflow step 4 (uvx ruff) and step 5 (uv run) rely on external package availability.
[PROMPT_INJECTION] (LOW): As a tool that processes external data and API responses, it provides a surface for indirect prompt injection. The skill includes instructions to mitigate this risk through input validation.
Evidence: Safety practices in SKILL.md explicitly mention validating API responses and external inputs.

python-script