The Agent Skills Directory

[SAFE]: The skill serves as a legitimate technical guide and developer tool for the Model Context Protocol ecosystem.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch documentation from official sources, specifically modelcontextprotocol.io and its associated GitHub repositories. These are recognized as trusted sources for the protocol's documentation and SDKs.
[COMMAND_EXECUTION]: The evaluation scripts (scripts/evaluation.py and scripts/connections.py) allow for the execution of a local MCP server process using the stdio transport. This is a core, documented feature of the protocol intended for local tool integration. The command to be executed is provided by the user as a CLI argument.
[CREDENTIALS_SAFE]: The skill documentation and evaluation harness correctly implement secret management by recommending that users store API keys (such as ANTHROPIC_API_KEY or GITHUB_TOKEN) in environment variables rather than hardcoding them in scripts or configuration files.
[INDIRECT_PROMPT_INJECTION]: The evaluation framework processes an XML file (evaluation.xml) containing natural language questions. While this is an ingestion point for external data, it is the intended purpose of the test suite. The harness uses specific XML tags (<summary>, <feedback>, <response>) to maintain structural boundaries in the model's output.

mcp-builder