The Agent Skills Directory

[SAFE]: The skill is entirely composed of Markdown documentation and reference files providing expertise on the Model Context Protocol. It does not contain executable code beyond illustrative snippets for Python and TypeScript development.- [PROMPT_INJECTION]: Deterministic detectors identified phrases such as 'Ignore previous instructions' within the file 'references/mcp-security-vulnerabilities.md'. These are verified as false positives, as they are explicitly presented as hypothetical examples of attack vectors in a security educational context.- [EXTERNAL_DOWNLOADS]: The documentation references official installation commands for established libraries such as 'mcp', 'httpx', and official '@modelcontextprotocol' SDK packages from standard registries. These references are used solely for instructional purposes to guide developers in setting up their environments.- [CREDENTIALS_UNSAFE]: The skill provides examples of configuration files (e.g., .claude.json) and CLI commands. Any references to sensitive environment variables or API keys utilize generic placeholders (e.g., 'YOUR_KEY', 'ghp_...') and do not contain hardcoded secrets.

mcp-expert