The Agent Skills Directory

[SAFE]: The skill implements safety best practices by explicitly instructing the agent to block path traversal attempts using '../' in file path inputs.
[SAFE]: The instructions require mandatory input and call validation, including schema checks, before executing any tool handlers.
[SAFE]: The skill enforces protocol-safe logging practices to prevent data corruption, such as using stderr for logs in Node.js and the logging module in Python instead of standard output.
[SAFE]: External references are limited to official documentation for the Model Context Protocol (modelcontextprotocol.io), which is the primary subject of the skill.
[SAFE]: The skill follows least-privilege principles by instructing the agent to perform minimal code changes and separate confirmed facts from assumptions.

mcp-server