my-mcp-server

Provides a standardized MCP server implementation that handles protocol initialization, message routing, and connection lifecycle according to the Model Context Protocol specification. The server manages bidirectional communication channels between MCP clients (like Claude Desktop or other LLM applications) and exposes tools/resources through a declarative interface. Implements request-response patterns, error handling, and graceful shutdown mechanics required by the MCP specification.

Enables developers to declare tools with JSON Schema-based argument specifications, which are automatically validated and exposed to MCP clients. Tools are registered through a declarative interface that maps function implementations to tool metadata (name, description, input schema). The server handles schema validation of incoming tool invocations before passing arguments to the implementation, preventing malformed requests from reaching handler code.

Allows the server to expose static or dynamic resources (files, templates, documentation, data) that MCP clients can request and retrieve. Resources are registered with metadata (URI, MIME type, description) and the server handles content retrieval, caching, and streaming. Clients can discover available resources and request specific content, enabling use cases like providing context documents, configuration files, or generated content to LLM applications.

Enables the server to register reusable prompt templates that MCP clients can discover and invoke. Prompts are defined with metadata (name, description, arguments) and can include dynamic content generation based on input arguments. The server handles prompt instantiation, argument substitution, and returns the final prompt text to the client, enabling consistent prompt engineering across multiple LLM applications.

Implements MCP protocol handshake and capability negotiation between the server and connecting clients. During initialization, the server advertises its supported features (tools, resources, prompts, sampling capabilities) and the client declares its capabilities. This enables graceful degradation and ensures both parties understand what functionality is available, preventing requests for unsupported features.

Provides standardized error handling that converts exceptions and failures into MCP-compliant error responses. When tool invocations, resource requests, or other operations fail, the server catches exceptions, formats them according to the MCP protocol (including error codes and messages), and returns them to the client without crashing. This ensures robust communication and enables clients to handle errors gracefully.

Enables the server to request that connected MCP clients (which have access to LLM models) perform sampling or inference on behalf of the server. The server can send sampling requests with prompts, model parameters (temperature, max tokens), and system instructions, and the client returns generated text. This allows MCP servers to leverage LLM capabilities without directly calling model APIs, enabling agentic workflows where the server orchestrates LLM calls.

Provides logging infrastructure to track MCP protocol messages, tool invocations, resource requests, and errors. The server logs incoming requests, outgoing responses, and internal state changes, enabling developers to debug integration issues and monitor server behavior. Logging can be configured at different verbosity levels to balance detail with performance.