ragcode-sse

This skill provides a robust architectural guide for integrating AI agents with the RagCode MCP (Model Context Protocol) server using the native Server-Sent Events (SSE) transport layer. It is designed for developers and AI engineers who need to bypass traditional configuration requirements and establish a direct, lightweight HTTP communication stream between their agents and the RagCode infrastructure. By implementing the SSE handshake and JSON-RPC messaging sequence, users can achieve seamless tool invocation and data retrieval, ensuring high performance and reliable state management without the overhead of additional daemon binaries or local configuration files.

• Enables direct connection via HTTP GET and POST methods using the standard /sse endpoint for real-time bi-directional messaging.

• Provides a verified, step-by-step handshake procedure including 'initialize' and 'notifications/initialized' methods.

• Includes comprehensive code examples for Bash (curl), Python (with sseclient), and generic JSON-RPC payloads for tool execution.

• Facilitates granular tool discovery through 'tools/list' and input schema validation for precise code search operations.

• Simplifies error handling and session management by utilizing the 'endpoint' event to dynamically extract the session-specific POST URL.

• Ensure the 'Accept: text/event-stream' header is consistently provided to maintain the persistent SSE connection.

• Always perform the full MCP handshake sequence before attempting any 'tools/call' operations to avoid session state errors.

• Match responses to requests using unique 'id' fields in the JSON-RPC payload to ensure thread-safe asynchronous communication.

• The SSE connection must remain open throughout the lifecycle of the task to receive asynchronous event updates from the server.

• Useful for automated workspace indexing, code retrieval, and agentic workflows where low-latency communication with RagCode is required.