openapi-atlas-mcp
Aggregates multiple OpenAPI specifications into a single searchable MCP server, enabling AI agents to query API endpoints across all services using natural language.
README
openapi-atlas-mcp
openapi-atlas-mcp aggregates multiple OpenAPI/Swagger specifications from your
backend services into a single searchable, documentation-focused
Model Context Protocol server. It lets an AI
agent (Claude Desktop, Cursor, Claude Code, VS Code) answer questions like
"show me the payload for submit order" across all of your configured
services in one query — instead of making you hunt through N separate Swagger UIs.
Install
# pip
pip install openapi-atlas-mcp
# uv (recommended for MCP servers)
uv tool install openapi-atlas-mcp
Quick start
- Create a config file (
.openapi-atlas.json) in your project root:
openapi-atlas-mcp init
This launches an interactive wizard. Or write the file manually:
{
"services": [
{
"name": "orders",
"url": "http://localhost:8000"
},
{
"name": "billing",
"url": "https://billing.internal.example.com",
"auth": {
"type": "bearer",
"token_env": "BILLING_API_TOKEN"
}
}
]
}
- Set any required environment variables:
export BILLING_API_TOKEN=your-secret-token
- Start the server:
openapi-atlas-mcp serve
Agent-host integration
Claude Desktop (claude_desktop_config.json)
{
"mcpServers": {
"openapi-atlas": {
"command": "uv",
"args": ["tool", "run", "openapi-atlas-mcp", "serve"]
}
}
}
Cursor (.cursor/mcp.json)
{
"mcpServers": {
"openapi-atlas": {
"command": "openapi-atlas-mcp",
"args": ["serve"]
}
}
}
Claude Code
claude mcp add openapi-atlas -- openapi-atlas-mcp serve
VS Code
Add to .vscode/mcp.json:
{
"servers": {
"openapi-atlas": {
"command": "openapi-atlas-mcp",
"args": ["serve"]
}
}
}
Configuration reference
The config file (.openapi-atlas.json or path via OPENAPI_ATLAS_CONFIG env var):
| Field | Type | Required | Description |
|---|---|---|---|
services |
ServiceConfig[] |
Yes | One or more backend services to aggregate |
auth |
AuthConfig | null |
No | Global default auth (overridable per-service) |
allow_insecure_http_hosts |
string[] |
No | Hosts allowed for plain HTTP (e.g. ["localhost"]) |
cache.spec_ttl_seconds |
int |
No | Spec cache TTL in seconds (default: 60) |
cache.disk_cache_dir |
string | null |
No | Custom disk cache directory |
ServiceConfig
| Field | Type | Required | Description |
|---|---|---|---|
name |
string |
Yes | Unique service identifier |
url |
string |
Yes | Base URL of the service (must include scheme) |
spec_path |
string | null |
No | Explicit spec path (e.g. /api/openapi.json). Auto-discovered if omitted. |
auth |
AuthConfig | null |
No | Per-service auth override |
AuthConfig
Secrets are referenced by env-var name — never inline. See docs/auth.md for patterns.
| Type | Required fields | Behavior |
|---|---|---|
none |
(none) | No auth headers sent |
bearer |
token_env |
Sends Authorization: Bearer <env value> |
basic |
username_env, password_env |
Sends Authorization: Basic <base64> |
headers |
headers_env |
Sends custom headers from a JSON env var |
Tools
| Tool | Description |
|---|---|
search_endpoints |
Keyword-search operations across all services. Returns ranked matches with scores. |
get_endpoint_detail |
Returns the fully $ref-resolved schema for one endpoint. |
list_services |
Lists all configured services with health, version, and freshness metadata. |
refresh_specs |
Force a fresh fetch of one or all services. Clears caches. |
When things go wrong
| Symptom | Likely cause | Fix |
|---|---|---|
ConfigNotFoundError |
No .openapi-atlas.json found |
Run openapi-atlas-mcp init or set OPENAPI_ATLAS_CONFIG |
MissingEnvVarError |
Referenced env var not set | export BILLING_API_TOKEN=... |
InsecureHttpHostError |
Plain HTTP to non-localhost | Add host to allow_insecure_http_hosts |
Service shows unhealthy |
Service unreachable | Check URL, auth, and network connectivity |
Service shows degraded |
Serving stale cache | Service was up before; check if it restarted |
| Empty search results | No services indexed yet | Call refresh_specs or check service health |
See docs/errors.md for the full warning catalog.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.