The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses npx to dynamically fetch and execute the @smithery/cli package from the Smithery registry.
[COMMAND_EXECUTION]: Employs shell commands including npx and rg (ripgrep) for environment discovery and automated installation procedures.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: Gathers MCP server definitions and configuration objects from the agent's context in SKILL.md.
Boundary markers: No explicit delimiters or instructions are provided to the agent to differentiate between trusted and potentially malicious server configurations provided in the prompt.
Capability inventory: The skill possesses file-write capabilities across multiple sensitive application directories and can execute arbitrary shell commands via npx (SKILL.md, references/clients.md).
Sanitization: Includes a manual verification step for secrets (API keys, tokens) before proceeding, but lacks validation for the executable commands or arguments within the server definitions.
[DYNAMIC_EXECUTION]: Programmatically generates JSON configuration files containing shell commands and arguments that are later executed by third-party clients (such as Claude Desktop or Cursor) during normal operation.
[PERSISTENCE]: Modifies persistent application configuration files (e.g., claude_desktop_config.json, settings.json) to register new servers that are automatically launched by the host application.

mcp-installer