The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use various uv CLI commands for environment management and script execution, such as uv init, uv sync, and uv run.
[EXTERNAL_DOWNLOADS]: The instructions describe how to download and install Python packages and tools from public registries using uv add and uv tool install.
[REMOTE_CODE_EXECUTION]: The skill promotes the use of uvx and uv tool install which fetch and execute external packages; however, these are standard package management functions interacting with trusted registries and are considered safe within the intended use-case.
[DYNAMIC_EXECUTION]: Provides methods for executing Python code snippets and scripts at runtime using uv run and uvx, which involves dynamic interpretation of Python source code as part of its primary functionality.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an environment for processing and executing user-supplied Python code, which constitutes an indirect injection surface. 1. Ingestion points: User-provided script files and inline code snippets passed to uv run via stdin. 2. Boundary markers: The skill does not explicitly define delimiters for untrusted content, relying on standard shell piping or heredoc structures. 3. Capability inventory: Subprocess execution for uv commands, including script execution, environment creation, and network-based package retrieval. 4. Sanitization: No specific sanitization or validation of the Python code logic is described, as the skill is intended for standard development tasks.

uv-python