h3-mcp-tools is a minimal MCP server built with H3 v2 that exposes a set of tooling endpoints for JSON-RPC based tool management. It provides a simple way to register tools, validate inputs against schemas, call tools, and list available tools. The package also supports dynamic and static resources, as well as streaming responses via a ReadableStream utility. To get started, install and run the MCP server, then interact with the /mcp endpoint to call tools (e.g., tools/call) or fetch the list of registered tools (e.g., tools/list). The included example demonstrates an echo tool and shows how to request tool calls and listings through standard JSON-RPC payloads. For streaming scenarios, you can use the createMcpStream helper to stream partial results before sending a final JSON-RPC response.

Prerequisites:

• Node.js (recommended LTS) and npm or pnpm/yarn
• Basic familiarity with npm-based packages

Installation steps:

1. Initialize a project (if needed) and install the package: npm install h3-mcp-tools

   or using your preferred package manager

2. Run the MCP server (example using npx): npx -y h3-mcp-tools

3. (Optional) If you plan to integrate into an existing H3 server, refer to the H3MCP class usage in the README and import the package where needed.

Notes:

• The server exposes a JSON-RPC API at the /mcp endpoint by default. Ensure your environment exposes this path if deploying behind a reverse proxy.

Tips and considerations:

• This MCP server is experimental and under active development. It aims to remain lightweight with minimal dependencies while demonstrating H3 v2 integration.
• Tools are registered with schemas for input validation using standard-schema based validators. You can also provide a jsonSchema for libraries supported by standard-json.
• Streaming responses are supported via ReadableStream using the createMcpStream helper. If you include finalResponse in your stream options, you can close the stream with a complete JSON-RPC response.
• Resources support both static definitions and dynamic handlers for list/read operations. You can batch define resources as well.
• If you need more capabilities, consider the official MCP TS SDK or TMCP for extended features.

Environment variables and config options: you can customize tool definitions, resource endpoints, and streaming behavior through the H3MCP setup. If you deploy in a container or cloud environment, ensure port exposure and route configuration for the /mcp endpoint.