MCP Integration
OnboardingHub provides a Model Context Protocol (MCP) server that lets AI assistants interact with your workspace. Use it with Claude Desktop, Cursor, or any MCP-compatible client.
Connection setup
Endpoint
POST https://onboarding-hub.com/mcp
The MCP server accepts JSON-RPC 2.0 requests over HTTP and uses the same Bearer token authentication as the REST API.
Claude Desktop configuration
Add this to your Claude Desktop claude_desktop_config.json:
{
"mcpServers": {
"onboardinghub": {
"url": "https://onboarding-hub.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_ACCESS_TOKEN"
}
}
}
}
Other MCP clients
Any MCP client that supports HTTP transport can connect using:
- URL:
https://onboarding-hub.com/mcp - Method:
POST - Headers:
Authorization: Bearer YOUR_ACCESS_TOKEN - Content-Type:
application/json
Authentication
The MCP server uses the same authentication as the REST API. Create an OAuth application in your workspace's Integrations > Developer or use an OAuth 2.0 access token.
The MCP server requires the api-access entitlement on your workspace's plan.
Available tools
Tools are filtered based on your token's scopes. Only tools whose required scope is present on your token are exposed.
Contact tools
| Tool | Description | Required scope |
|---|---|---|
list_contacts |
List contacts with filtering and pagination | contacts:read |
get_contact |
Get a contact by ID | contacts:read |
create_contact |
Create a new contact | contacts:write |
update_contact |
Update an existing contact | contacts:write |
Organisation tools
| Tool | Description | Required scope |
|---|---|---|
list_organisations |
List organisations with filtering | organisations:read |
get_organisation |
Get an organisation by ID | organisations:read |
create_organisation |
Create a new organisation | organisations:write |
update_organisation |
Update an existing organisation | organisations:write |
Enrollment tools
| Tool | Description | Required scope |
|---|---|---|
list_enrollments |
List enrollments with filtering | enrollments:read |
get_enrollment |
Get an enrollment by ID | enrollments:read |
create_enrollment |
Enroll a contact in a guide | enrollments:write |
update_enrollment |
Update enrollment status | enrollments:write |
Guide tools
| Tool | Description | Required scope |
|---|---|---|
list_guides |
List published guides | guides:read |
get_guide |
Get a guide with sections and steps | guides:read |
Webhook tools
| Tool | Description | Required scope |
|---|---|---|
list_webhook_endpoints |
List webhook endpoints | webhooks:manage |
create_webhook_endpoint |
Create a webhook endpoint | webhooks:manage |
update_webhook_endpoint |
Update a webhook endpoint | webhooks:manage |
Resources
The MCP server also exposes published guides as readable resources. With the guides:read scope, Claude and other AI assistants can read the full content of your guides (including section descriptions and step content).
Resources use the URI format: onboardinghub://guides/{guide_id}
Example usage
Once connected, you can ask your AI assistant things like:
- "List all contacts in the Acme Corp organisation"
- "Create a new contact with email [email protected]"
- "Enroll contact abc123 in guide xyz789"
- "Show me the enrollment progress for [email protected]"
- "What guides do I have published?"
- "Read the Getting Started guide"
- "Create a webhook endpoint for enrollment.completed events"
Protocol details
- Protocol: JSON-RPC 2.0 over HTTP
- Transport: Stateless HTTP (no server-side sessions)
- Server name:
onboardinghub - Version:
1.0.0 - The
GET /mcpendpoint returns405 Method Not Allowed(SSE streaming is not supported in stateless mode) - The
DELETE /mcpendpoint returns204 No Content(no sessions to clean up)
Next steps
- Authentication -- create an API key for MCP access
- Endpoints reference -- the same operations are available via REST