The Agent Skills Directory

[SAFE]: The skill consists of instructional documentation for software development without any embedded malicious code or suspicious override instructions.
[EXTERNAL_DOWNLOADS]: The skill fetches technical documentation from official Model Context Protocol repositories on GitHub and the protocol's main website. These are well-known, legitimate sources directly relevant to the skill's purpose.
[COMMAND_EXECUTION]: Development commands like npm run build and python -m py_compile are mentioned as part of the build and verification process for the server code being developed.
[INDIRECT_PROMPT_INJECTION]: The skill processes external documentation, which is a potential ingestion point for untrusted data. However, the risk is mitigated as the skill targets official protocol documentation and explicitly instructs users to implement sanitization and validation.

mcp-builder