The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill describes an MCP server that processes data from Umbraco instances, which acts as a surface for indirect instructions. Ingestion points: External data fetched from Umbraco API endpoints via tool calls. Boundary markers: Not explicitly defined in instructions, though SDK-level sanitization is noted. Capability inventory: The server exposes tools for reading, searching, and mutating Umbraco content and media. Sanitization: Documented automatic validation includes rejection of control characters, path traversal (../), and UUID format verification.
[COMMAND_EXECUTION]: The local development guide contains instructions for shell commands to build and link the project. Evidence: Commands such as 'npm run build', 'npm link', 'rm -rf node_modules', and 'ln -s' are described in local-dev-testing.md for environment setup.
[EXTERNAL_DOWNLOADS]: The documentation guides the agent and user to execute '@umbraco-cms/mcp-dev' using npx, which fetches the package from the official NPM registry to initiate the server.

mcp-cli