Turn any REST API into an MCP server
Import a Postman collection or Swagger/OpenAPI spec, test the endpoints, and publish them as tools your AI assistant can call — in minutes, not sprints.
No credit card needed to start.
From API to AI tool in three steps
No SDKs, no server to deploy, no protocol code to maintain.
-
1
Bring your API
Upload a Postman collection (with its environment) or a Swagger/OpenAPI spec, paste a spec URL, or add endpoints by hand.
-
2
Test and publish
Try each endpoint with real requests. Safe GET endpoints that respond are published automatically; the rest wait for your OK.
-
3
Connect your assistant
Group endpoints into an MCP server and paste its URL into your AI client. Every endpoint becomes a tool it can call.
Everything you need to ship MCP tools
Built for developers who already have an API and want AI assistants to use it.
Postman & Swagger import
Folders, collection variables, environments, inherited auth, path parameters and request bodies all come across. Whole Postman workspaces too.
Built-in request tester
Send real requests with parameters, headers, auth and a raw body before anything goes live.
OAuth 2.0 protected servers
Require sign-in before an assistant can use your tools, so only people you authorize get access. Paid plans.
MCP call logs
See every tool call: arguments, status, timing and response, so you can debug what the assistant actually did.
Failure alerts
Get notified when a tool call fails, an import finishes, or something needs your attention.
Groups & multiple servers
Organize endpoints into groups and attach them to separate MCP servers, each with its own URL — one per assistant, team or use case.
Secure by default
Your API is only exposed the way you choose.
- Passwords hashed with PBKDF2; access tokens stored only as SHA-256 hashes
- Requests to localhost and private network addresses are blocked
- Per-server OAuth 2.0 with PKCE for protected MCP servers
- Endpoints stay as drafts until they're tested or you publish them
Simple, transparent pricing
Start with a 7-day free trial. Upgrade when you need more.
Free
Try it out, no cost
- 2 MCP servers
- 2 groups
- 5 endpoints
- 50 MCP calls during the trial
- 1 user
- OAuth-protected servers
- Community support
Starter
For solo builders
- 5 MCP servers
- 5 groups
- 15 endpoints
- 5,000 MCP calls / month
- 3 users
- OAuth-protected servers
- Email support
Pro
For scaling products
- 50 MCP servers
- 50 groups
- 200 endpoints
- 50,000 MCP calls / month
- 25 users
- OAuth-protected servers
- Priority email support
Business
For established businesses
- 100 MCP servers
- 1,000 groups
- 4,000 endpoints
- 100,000 MCP calls / month
- 75 users
- OAuth-protected servers
- Priority + chat support
For large organizations
Run MCP Generator across your whole organisation with no caps on MCP servers, groups, endpoints, MCP calls or users. Every server can be protected with OAuth sign-in, and a dedicated account manager works with your team on onboarding, rollout and custom terms.
- Unlimited MCP servers, groups and endpoints
- Unlimited MCP calls every month
- Unlimited users
- OAuth-protected servers
- Dedicated account manager
- Custom pricing and invoicing
Prices in USD, billed monthly. Change or cancel your plan any time from Billing.
Frequently asked questions
What is MCP?
The Model Context Protocol is an open standard that lets AI assistants call external tools. MCP Generator turns your REST endpoints into MCP tools, so an assistant can look up, create or update data in your system when you ask it to.
Do I need to write any code?
No. Import a Postman collection or Swagger/OpenAPI spec (or fill in a form for each endpoint), test it, and publish. MCP Generator hosts the MCP server for you.
Which AI assistants does it work with?
Any MCP client that can connect to a remote MCP server by URL. You copy the server's URL from MCP Generator into your client's MCP settings.
What happens after the free trial?
The trial lasts 7 days and includes 50 MCP calls. After that, your MCP servers are paused until you choose a paid plan. Your servers, groups and endpoints are kept, and everything resumes as soon as you upgrade.
What if I hit my plan's limits?
You'll be asked to upgrade before creating more MCP servers, groups or endpoints than your plan allows. If you use up your plan's MCP calls, further tool calls are refused until your plan renews or you upgrade.
Can my API's private hosts be reached?
No. For safety, requests to localhost and private network IP addresses are blocked. Your API must be reachable over the public internet.
Put your API in front of AI assistants today
Import your first collection in minutes.
Start your 7-day free trial