The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions and examples include commands to download documentation and reference guides from github.com/anthropics and github.com/modelcontextprotocol. These are established and trusted sources for MCP development and the skill's own origin repository.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external API specifications (REST/GraphQL) to generate MCP server implementations. This constitutes a potential attack surface where maliciously crafted API documentation could attempt to influence the agent's code generation. The skill mitigates this by mandating human verification, a dedicated security review phase, and read-only evaluation requirements.- [COMMAND_EXECUTION]: The skill utilizes local scripts (validate-evals.mjs) and system tools (Bash) for validation and implementation tasks. These are standard operations for a development-oriented skill and do not involve piped remote execution.

mcp-builder