The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch documentation and SDK information from official Model Context Protocol sources on GitHub and modelcontextprotocol.io. These downloads are part of the intended educational and development workflow.- [COMMAND_EXECUTION]: The evaluation harness (scripts/evaluation.py) utilizes a command-line interface to launch MCP servers as local subprocesses for testing. This capability is required for the tool's primary purpose of server verification.- [DATA_EXFILTRATION]: The skill performs network requests to the Anthropic API to conduct evaluations. It properly advises the use of environment variables for storing the ANTHROPIC_API_KEY, adhering to standard security practices.- [PROMPT_INJECTION]: The evaluation process involves reading questions from local XML files and feeding them to an LLM. While this is a form of untrusted data ingestion, the script implements structure-based response parsing to minimize the risk of the agent deviating from its instructions.

mcp-builder