The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for executing Python scripts and various CLI tools (e.g., ruff, black, pytest, uvicorn) using uv run and uvx. This is the intended functionality of the tool for managing Python environments and running development workflows.
[EXTERNAL_DOWNLOADS]: Instructions include examples of adding dependencies from remote Git repositories (uv add git+https://...) and alternative package indexes. These are standard features of package managers for retrieving code from external sources.
[INDIRECT_PROMPT_INJECTION]: The skill enables the agent to process and execute code from Python scripts and project configuration files (pyproject.toml). If an attacker provides malicious files, they could potentially influence the agent's behavior during the execution of UV commands.
Ingestion points: Python scripts (.py) and project metadata files (pyproject.toml, uv.lock) provided by the user or fetched from remote repositories.
Boundary markers: The instructions do not specify explicit delimiters or warnings for the agent to ignore instructions embedded in the processed data.
Capability inventory: Full execution of Python code and shell commands via the uv toolset across all documented workflows.
Sanitization: The documentation does not describe specific sanitization or validation steps for external content beyond standard UV behavior.

using-uv