OpenAPI to MCP
Import a Swagger/OpenAPI spec, generate MCP tools, edit descriptions for AI agents, and publish selected operations.
[Explore API publishing](/openapi-to-mcp)
Import a Swagger/OpenAPI spec or connect a PostgreSQL or SQL Server scope, generate MCP tools, edit descriptions for agent behavior, and test the result instantly in Gateway Chat. The Gateway keeps auth, validation, limits, and logs server-side.
Step 1
Paste the OpenAPI URL, add upstream auth, and let the Gateway discover operations.
OpenAPI URL
https://demoapi.../swagger.json
Base URL
/api/v1
Auth
Bearer Token
Detected
18 operations ready for review
Gateway output
MCP surface created
MCP URL
/mcp/service-ops
Credentials stay server-side.
Step 2
Pick which operations agents can see and rewrite descriptions so tool choice is obvious.
Generated tools
search_orders
create_ticket
get_customer_status
Selected tool
search_orders
Agent-ready description
Search approved orders by customer, status, or date range before support actions.
Only published tools appear in MCP clients.
Gateway Chat
Ask a real question. Chat selects the published MCP tool and shows the call.
Which open Acme orders need support follow-up?
I will call the published order tool and summarize the result.
search_orders
succeeded · 284ms3 open orders found. 1 delayed shipment needs follow-up.
Step 4
After tool calls run, usage shows call volume, success rate, latency, and redacted logs.
Total Calls
1,248
Success
98.7%
Avg Latency
284ms
Last 24h
312
Call Volume
Tool Analytics
request arguments redacted · response stored
What you can publish
Generate a useful MCP catalog, curate what agents can see, and keep execution controls in the Gateway.
Import a Swagger/OpenAPI spec, generate MCP tools, edit descriptions for AI agents, and publish selected operations. Publish PostgreSQL or SQL Server scopes as metadata-only schema tools plus validated QueryPlan or SQL executors.
Keep credentials, validation, limits, quotas, and redacted logs in the Gateway instead of prompts or local client config.
Create a safe MCP surface without hand-written adapters
Start with a Swagger/OpenAPI URL or a PostgreSQL or SQL Server database connection.
Create a model-facing MCP catalog without writing a custom MCP server for every source.
Review, rename, hide, and describe tools so AI clients choose the right capability with better arguments.
Validate the surface in Gateway Chat, then connect Claude, Cursor, VS Code, or any MCP-compatible client.
Product Walkthrough
Watch the product flow from source import to generated tools, curated descriptions, published MCP URL, and in-app testing.
Pricing
Start free, then scale capacity as MCP traffic increases.
For trying the gateway and small projects. $0forever
For production workflows and growing usage. $9per month
or $69 per year
For organizations with high throughput. $199per month
or $499 per year
Start with an OpenAPI spec, connect a database scope, tune the tools, and test them in Gateway Chat before connecting external clients.