The Agent Skills Directory

[SAFE]: The skill consists of reference documents and procedures for developers building MCP servers. It contains no executable code or malicious instructions directed at the agent or the host environment.
[EXTERNAL_DOWNLOADS]: The documentation references official tools and libraries from the Model Context Protocol project and Anthropic, such as @modelcontextprotocol/sdk and @modelcontextprotocol/inspector. These are well-known and trusted resources for the intended development workflow.
[COMMAND_EXECUTION]: The skill outlines the use of standard development tools like npx, node, and brew for local development and testing. These commands are provided as guidance for the developer to follow manually and are not intended for autonomous execution by the agent. The references/security.md file also mentions sensitive file paths (e.g., ~/.ssh/id_rsa) purely in an educational context to describe external threat vectors.

mcp-server