openapi.city
Sign in
90% → 3% context window usage

Your AI agent is
guessing APIs. Stop it.

Give your AI agent precise, resolved OpenAPI context — a single endpoint, a compact summary, or a schema. Via MCP or REST. 100+ APIs indexed.

Tell your agent:

Sign up for openapi.city. Read https://openapi.city/signup.md for instructions.

— Or sign up yourself the old-school way —

Get API Key → Read the docs
Without openapi.city CTX 0%
With openapi.city CTX 0%
105
API providers indexed
16,933
Endpoints parsed
97%
Smaller responses
How it works
Zero-prep API integration for agents

Your AI agent queries openapi.city and gets exactly what it needs — a single endpoint, a schema, or a compact summary. No manual spec hunting.

01

Find an API

Search across 100+ indexed providers. Filter by category, tag, or keyword. Get structured results instantly.

02

Get what you need

Request a single endpoint, a tag group, or an AI-optimized summary. Nested schemas available on-demand via get_schema — lean responses, no bloat.

03

Ship the integration

Your agent has auth requirements, parameters, request/response schemas, and base URLs. Enough context to write working code.

Features
Stop burning tokens on full API specs

Every feature designed to minimize context window waste and maximize agent autonomy.

⚡

Smart extraction

Pull a single endpoint with resolved schemas, or get a compact summary that fits in a system prompt. No more 100KB spec dumps.

🔗

MCP protocol native

Full MCP Streamable HTTP support. Tools, resources, and prompts — your agent connects and starts working immediately.

🔍

Full-text search

Search across providers, endpoints, and descriptions. Find the right API operation in milliseconds.

🔄

Always current

Weekly automated refreshes. Swagger 2.0 auto-converted to OpenAPI 3.0. Version history tracked.

🛡

Resolve any spec URL

POST /resolve with a URL. We fetch, validate, parse, and index it. Your agent can integrate with APIs we haven't indexed yet.

📊

Usage-based billing

100 calls/month free. €10/mo for 5,000. Simple, transparent, no surprises. Stripe-powered.

Code examples
Two protocols, one API key
REST API GET
# Get a single endpoint with resolved schemas

GET /api/v1/providers/stripe/endpoints/
    POST/charges

Authorization: Bearer oac_live_...

# Response: parameters, request body,
# response schemas (refs to schemas)
MCP Server POST
# List Slack endpoints via MCP tool

{
  "name": "list_endpoints",
  "arguments": {
    "provider": "slack",
    "tag": "chat"
  }
}

# Returns structured endpoint list
Pricing
Start free, scale when ready

No credit card required. Upgrade when your agents need more.

Free
€0 /month

For hobby projects and exploration

  • ✓ 100 API calls / month
  • ✓ Full REST API access
  • ✓ Full MCP server access
  • ✓ 100+ indexed APIs
  • ✓ Community support
Get Started
POPULAR
Pro
€10 /month

For production AI agents

  • ✓ 5,000 API calls / month
  • ✓ Full REST API access
  • ✓ Full MCP server access
  • ✓ 100+ indexed APIs
  • ✓ €10 per 1,000 overage calls
  • ✓ Custom API providers
  • ✓ Priority support
Get Started
FAQ
Common questions
What is openapi.city?
An API spec index and delivery service built for AI agents. We parse, resolve, and host OpenAPI specifications so your agent can query exactly what it needs — a single endpoint, a compact summary, or a full schema — without dumping entire specs into the context window.
How does it work with my AI agent?
Your agent calls openapi.city via MCP or REST, gets back structured spec data, and uses it to write correct API integrations. No more guessing parameter names or reading 100KB spec files.
What's the difference between REST API and MCP?
Same data, different protocol. Use the REST API for scripted pipelines and direct HTTP calls. Use the MCP server for agent-native tools — Claude, Cursor, Windsurf — where the agent discovers and calls tools automatically.
Which AI tools are supported?
Anything that speaks MCP or HTTP. That includes Claude Desktop, Cursor, Windsurf, Copilot, OpenClaw, and custom agents. If it can call an API, it can use openapi.city.
How do schema references work?
OpenAPI specs use $ref to reference other schemas. We keep responses lean by preserving refs — call get_schema to fetch any nested schema on-demand. Smaller payloads, faster responses.
Can I add my own API specs?
Yes — on the Pro plan. POST /resolve with a spec URL and we fetch, parse, and index it for your agents. Upgrade to Pro →
How are specs kept up to date?
Weekly automated refreshes from source URLs. Swagger 2.0 specs are auto-converted to OpenAPI 3.0. Version history is tracked.
What happens if I exceed my monthly calls?
Pro: €10 per 1,000 overage calls, billed automatically. Free: requests return 429 with a clear message. No surprise charges on either plan.
Do I need a credit card?
No. The free tier (100 calls/month) starts immediately. Just tell your agent:
Sign up for openapi.city. Read https://openapi.city/signup.md for instructions.
The API reference
your agent was missing.

100 free calls. No credit card. Start building in 2 minutes.

Get Your API Key → Read the docs