> ## Documentation Index
> Fetch the complete documentation index at: /docs/llms.txt
> Use this file to discover all available pages before exploring further.

# OpenAPI Import

> Generate MCP servers from OpenAPI specifications

## Overview

Paste an OpenAPI spec URL and BoostGPT automatically generates a working MCP server. No coding required.

## Quick Start

**1. Get OpenAPI URL**

Find your API's OpenAPI specification URL:

```text theme={null}
https://api.yourservice.com/openapi.json
```

**2. Create Server**

* Go to **Dashboard -> MCP Servers**
* Click **Create Server**
* Select **Import OpenAPI**

**3. Paste URL**

Paste your OpenAPI spec URL and click **Generate**

**4. Server Ready**

BoostGPT generates your MCP server instantly:

```text theme={null}
https://mcp.your-server-id.boostgpt.co
```

**5. Connect to your AI teammate**

* AI teammate -> **Tools** tab -> **Add Tool** -> **Add New**
* Enter your MCP URL
* Start using it in the builder chat

## OpenAPI Format

Your OpenAPI spec should be in JSON or YAML format:

```yaml theme={null}
openapi: 3.0.0
info:
  title: My API
  version: 1.0.0
servers:
  - url: https://api.example.com/v1
paths:
  /users:
    get:
      summary: List users
      parameters:
        - name: limit
          in: query
          schema:
            type: integer
      responses:
        '200':
          description: Success
  /users:
    post:
      summary: Create user
      requestBody:
        content:
          application/json:
            schema:
              type: object
              properties:
                name:
                  type: string
                email:
                  type: string
```

## Auto-Generated Tools

BoostGPT converts each API endpoint into an AI teammate tool:

| API Endpoint | Generated Tool | Description |
| - | - | - |
| `GET /users` | `list_users` | List users with optional limit |
| `POST /users` | `create_user` | Create a new user |
| `GET /users/{id}` | `get_user` | Get user by ID |
| `PUT /users/{id}` | `update_user` | Update user information |
| `DELETE /users/{id}` | `delete_user` | Delete user |

## Example: CRM API

**OpenAPI Spec:**

```yaml theme={null}
openapi: 3.0.0
info:
  title: CRM API
  version: 1.0.0
paths:
  /contacts:
    post:
      summary: Create contact
      requestBody:
        required: true
        content:
          application/json:
            schema:
              properties:
                name:
                  type: string
                email:
                  type: string
                company:
                  type: string
  /deals:
    get:
      summary: List deals
      parameters:
        - name: status
          in: query
          schema:
            type: string
```

**Generated Tools:**

```text theme={null}
Teammate can now use:

1. create_contact
   - Parameters: name, email, company
   - Creates a new CRM contact

2. list_deals
   - Parameters: status (optional)
   - Lists deals filtered by status
```

## Authentication

Add authentication to your OpenAPI spec:

```yaml theme={null}
components:
  securitySchemes:
    ApiKeyAuth:
      type: apiKey
      in: header
      name: X-API-Key
    BearerAuth:
      type: http
      scheme: bearer
security:
  - ApiKeyAuth: []
```

Configure credentials in **Dashboard -> MCP Servers -> Your Server -> Settings**

## Testing

Test your MCP server after creation:

**1. Add to your AI teammate**

Connect your MCP server to any AI teammate

**2. Open the builder chat**

Open a Site and use the builder chat

**3. Test Tools**

```text theme={null}
User: "List all users in the system"
Teammate: *uses list_users tool* "Found 25 users..."

User: "Create a contact named John Doe"
Teammate: *uses create_contact tool* "Contact created successfully"
```

## Common Issues

> [accordion] Invalid OpenAPI URL
>
> Ensure your URL returns valid OpenAPI 3.0+ JSON or YAML. Test it in a browser first.

> [accordion] Authentication failing
>
> Add security schemes to your OpenAPI spec and configure credentials in MCP server settings.

> [accordion] Tools not working
>
> Check that your API endpoints return proper responses. Review error messages in AI teammate Insights.

## Best Practices

1. **Clear endpoint descriptions** - AI uses descriptions to understand when to call tools
2. **Include examples** - Add example requests/responses in your OpenAPI spec
3. **Document parameters** - Describe what each parameter does
4. **Use semantic naming** - Endpoint names like `/users` become tool names like `list_users`
5. **Version your API** - Use `/v1/` in paths to support multiple versions

## Public APIs

Many public APIs provide OpenAPI specs:

```text theme={null}
Stripe: https://raw.githubusercontent.com/stripe/openapi/master/openapi/spec3.json
GitHub: https://raw.githubusercontent.com/github/rest-api-description/main/descriptions/api.github.com/api.github.com.json
```

Import these to create custom connectors for public services.

## Next Steps

> [card] [Authentication](/docs/integrations/custom/authentication.md)
>
> Configure API authentication

> [card] [Postman Import](/docs/integrations/custom/postman.md)
>
> Alternative: Upload Postman collection


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.