MCP API Overview
The Unhook MCP (Model Context Protocol) server provides a standardized interface for AI assistants to access webhook data and debugging tools.Base URL
Authentication
All requests require authentication using a Bearer token:Transport
The MCP server supports HTTP Server-Sent Events (SSE) transport for real-time bidirectional communication.Protocol
The server implements the Model Context Protocol v1.0 specification with JSON-RPC 2.0 messages.Request Format
Response Format
Available Methods
Resources
method
Lists all available resourcesResponse:
method
Reads a specific resourceParameters:
uri(string, required): Resource URI to read
Tools
method
Lists all available toolsResponse includes:
search_events- Search webhook eventssearch_requests- Search webhook requestsanalyze_event- Analyze specific eventanalyze_request- Analyze specific requestget_webhook_stats- Get webhook statistics
method
Executes a tool with parametersParameters:
name(string, required): Tool namearguments(object, required): Tool-specific arguments
Prompts
method
Lists all available promptsResponse includes:
debug_webhook_issue- Debugging workflowanalyze_failures- Failure analysisperformance_report- Performance analysis
method
Gets a specific prompt templateParameters:
name(string, required): Prompt name
Rate Limits
- Requests per minute: 60
- Concurrent connections: 5
- Response size: 10MB max
Error Handling
Errors follow the JSON-RPC 2.0 error format:Common Error Codes
Example Session
SDK Support
The MCP server is compatible with:- @modelcontextprotocol/typescript-sdk
- Claude Desktop MCP client
- Cursor AI integration
Webhooks Integration
The MCP server automatically scopes data access to your organization based on the API key. You can only access webhooks, events, and requests that belong to your organization.Next Steps
- Review the Tool Reference for detailed tool documentation
- Check the Resource Reference for resource schemas
- See Integration Examples for common use cases