The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill specifies that the agent should manage Python projects by processing files such as pyproject.toml, uv.lock, and requirements.txt. These files represent an ingestion point for external data that could theoretically contain malicious instructions or influence agent logic. This is an inherent risk in development workflows.
Ingestion points: Project configuration files including pyproject.toml, uv.lock, requirements.txt, and .python-version.
Boundary markers: The skill does not define specific prompt delimiters for external file content.
Capability inventory: The skill enables subprocess execution via uv run and uvx, as well as network-enabled package management via uv add and uv sync.
Sanitization: No specific sanitization or validation logic for the content of these configuration files is mentioned.
[COMMAND_EXECUTION]: The skill defines a comprehensive command set using the uv utility for environment and package management. It incorporates security best practices by strictly forbidding sudo pip, sudo uv, and system-level installations, which effectively mitigates common vectors for system contamination and privilege escalation.

python