The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the GitHub CLI (gh) and Git to automate repository creation and initial commits (git init, gh repo create). These actions are explicitly described in the workflow and require user confirmation before execution.- [EXTERNAL_DOWNLOADS]: The generated project configuration (requirements.txt) specifies standard Python dependencies including fastmcp, starlette, uvicorn, and httpx. These are fetched from standard package registries (PyPI) during the user's deployment process.- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided themes to generate project metadata and tool stubs. It includes a specific safety instruction to treat the input theme as data rather than instructions to prevent prompt injection during the elicitation phase. Evidence: SKILL.md (Etapa 1 and Regras section).- [DATA_EXPOSURE]: The generated server.py includes a permissive CORS policy (allow_origins=["*"]) to facilitate connection from web-based AI clients (Gemini Spark, Claude Web, ChatGPT). While broad, this is the intended configuration for the described use case of a public remote MCP server.- [CREDENTIALS_SAFE]: The skill explicitly forbids hardcoding secrets or API keys in the generated code or documentation, instructing users to manage sensitive data through environment variables on the hosting platform (Vercel).

remote-mcp