The Model Context Protocol (MCP) has significantly improved how we integrate custom tools into AI development workflows like Cursor, Claude Code, and Windsurf. However, the current workflow introduces a repetitive bottleneck: every time you want an AI agent to interact with a new or custom REST API, you have to manually write Python or TypeScript SDK boilerplate to wrap those specific endpoints into compliant MCP tool schemas.
To eliminate this integration friction, I built a lightweight cloud gateway that automates the entire pipeline. It dynamically parses standard OpenAPI JSON specifications and provisions a hosted, remote MCP server running over Server-Sent Events (SSE) instantly.
The Core Problem
When building workflows with AI agents, exposing internal tools or external APIs usually means:
Writing boilerplate code to define tools, schemas, and execution blocks.
Redeploying or managing local configuration files whenever an API endpoint changes.
Handling manual protocol handshakes for simple HTTP requests.
This gateway acts as a dynamic translation layer, converting standard API documentation directly into machine-readable tool schemas on the fly.
How It Works
The setup requires no manual coding or local server management:
Submit the Spec: Provide a valid OpenAPI JSON specification URL or paste the raw JSON block directly into the gateway dashboard.
Generate the Endpoint: The gateway processes the paths, parameters, and payloads, generating a persistent, isolated remote server routing URL.
Connect to the Agent: Add the generated URL directly to your AI client settings (such as Cursor Features -> MCP) using the sse transport type. Your agent will instantly discover, inspect, and invoke the underlying API endpoints as tools.
Architecture Overview
The pipeline is built on an open, reliable stack designed for real-time tool routing:
FastAPI Backend: Handles the core application logic, parsing validation, and manages the active Server-Sent Events (SSE) streams required for real-time agent communication.
Supabase Integration: Manages persistent metadata indexing and secure, multi-account routing keys to keep configuration maps isolated.
Dynamic Protocol Mapping: Automatically translates complex OpenAPI request bodies, path parameters, and query variables into valid Model Context Protocol tool definitions without human intervention.
Project Links
Live Gateway Dashboard: https://mcp-bridge-saas.onrender.com
Current Constraints and Feedback
Please note that the service is currently hosted on Render's free tier. If the gateway dashboard or generated endpoints take a brief moment to respond on the first request, the container is simply spinning up from a standard 40-second cold start.
I am actively stress-testing the schema translation engine against highly nested data payloads and large-scale enterprise API definitions. If you are looking to expose your custom APIs to your agent workflows without writing manual wrapper code, please give the gateway a run. Any feedback regarding parsing edge cases, validation errors, or routing latency is highly appreciated.
Top comments (0)