The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download installation scripts from astral.sh and github.com/astral-sh/uv. These are the official domains and repositories for the Astral organization, which develops the uv tool. These references are documented neutrally as they originate from a well-known service provider.
[REMOTE_CODE_EXECUTION]: The skill documentation includes commands that download and immediately execute remote scripts, such as curl -LsSf https://astral.sh/uv/install.sh | sh and the PowerShell equivalent using iex. While these patterns involve remote code execution, they are the standard, documented installation methods for the tool and target the official vendor infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it instructions the agent to process potentially untrusted data from external files such as requirements.txt and pyproject.toml.
Ingestion points: Files are ingested via commands like uv add -r requirements.txt and uv sync (which reads pyproject.toml) as seen in SKILL.md.
Boundary markers: None are present; the tool processes the files according to standard package manager logic.
Capability inventory: The skill has the capability to perform network operations (package installation), write to the file system (uv init, uv lock), and execute arbitrary code via uv run as documented in both SKILL.md and references/advanced-patterns.md.
Sanitization: There is no explicit sanitization of the input file content; the skill relies on the underlying tool's resolution and validation mechanisms.

uv-package-manager