The Agent Skills Directory

[REMOTE_CODE_EXECUTION] (CRITICAL): In references/MANDATORY_PATTERNS.md, the skill includes instructions to install the uv tool using curl -LsSf https://astral.sh/uv/install.sh | sh. This is a piped execution of a remote script, which is a major security risk that allows for arbitrary code execution from an external source without verification.
[COMMAND_EXECUTION] (HIGH): The 'Skill Gateway' pattern in references/PROGRESSIVE_DISCLOSURE.md utilizes subprocess.run to execute local Python scripts based on paths and arguments passed through a params dictionary. This implementation lacks sanitization, enabling arbitrary command injection if the input parameters originate from an untrusted source or user prompt.
[EXTERNAL_DOWNLOADS] (MEDIUM): The skill's documentation retrieval workflow in SKILL.md and references/MANDATORY_PATTERNS.md relies on web_fetch to retrieve content from gofastmcp.com. As this domain is not a recognized trusted source, processing this content as authoritative documentation creates a surface for Indirect Prompt Injection (Category 8).
[DATA_EXFILTRATION] (LOW): The skill performs network requests to a non-whitelisted domain via web_fetch. While the stated purpose is documentation retrieval, any external network communication to non-trusted domains carries a inherent risk of data exposure.

creating-mcp-servers