Overview
An MCP server that provides a universal interface for AI model tools, enabling seamless integration with modern code assistants and IDEs such as Cursor and Claude Desktop. It acts as a bridge between Sunra.ai and your development environment, exposing Sunra's models and tools via the Model Context Protocol. The server supports listing available models, retrieving their schemas, submitting processing jobs, streaming results to clients, and other MCP capabilities as the protocol evolves. Security is handled by API keys managed through environment variables or runtime configuration, avoiding credential exposure in client code. Quick Start demonstrates installation and running with no local build required (for example, npx @sunra/mcp-server --transport http --port 3925). For Cursor IDE and Claude Desktop, you configure an MCP server entry pointing to the local HTTP endpoint or run in stdio mode, then access tools like list-models and model-schema. Comprehensive documentation is available in the mcp-server directory for full tool lists, development guidance, and troubleshooting.
Features
Universal MCP interface
Exposes Sunra's models and tools through the Model Context Protocol to enable integration with editors and IDEs.
Model discovery
Supports listing available models so clients can understand what is available.
Schema access
Fetch model schemas to understand inputs, outputs, and constraints.
Job submission and streaming
Submit processing jobs and stream results back to clients in real time.
Secure API key handling
API keys are managed via environment variables or runtime configuration to protect credentials.
Editor/IDE integrations
Plug-and-play with code assistants and IDEs such as Cursor and Claude Desktop.
Zero build requirement
No local build needed; run the server with a simple NPX command.
Glue-free setup
No glue code needed—run the server and connect your tool directly.
Who Is This For?
- Developers:Integrate Sunra models into editors and IDEs via the MCP server.
- Code editors/IDEs:Connect Cursor, Claude Desktop, and other tools to access models and run tasks through MCP.
- AI/ML engineers:List models, fetch schemas, submit jobs, and stream results within their workflows.
