Webhook MCP server
{
"mcpServers": {
"webhook": {
"type": "http",
"url": "https://mcp.webhook.co/mcp"
}
}
}{
"mcpServers": {
"webhook": {
"type": "http",
"url": "https://mcp.webhook.co/mcp"
}
}
}{
"mcpServers": {
"webhook": {
"type": "http",
"url": "https://mcp.webhook.co/mcp"
}
}
}{
"servers": {
"webhook": {
"type": "http",
"url": "https://mcp.webhook.co/mcp"
}
}
}{
"mcpServers": {
"webhook": {
"type": "http",
"url": "https://mcp.webhook.co/mcp"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 1.0.0
What it does
webhook.co is a service that provides a permanent, signed webhook URL for receiving and inspecting webhooks from external services. Developers use it to capture webhook events, verify their signatures, replay them to localhost, and forward them to other destinations—all accessible via CLI, API, dashboard, or MCP.
- Capture webhooks from services like Stripe, GitHub, or Shopify at a permanent URL, then inspect the full request and signature verification result.
- Forward captured webhooks to your local development server (http://localhost:3000) to test your webhook handler without deploying.
- Replay past webhook events to your local server or other destinations to debug and test error handling.
Tools this server exposes
| Tool | Does |
|---|---|
| endpoints_create | Create a new webhook endpoint and receive a permanent, signed ingest URL. |
| events_list | List captured webhook events for an endpoint with headers, body, and signature verification status. |
| events_get | Retrieve the full details of a single captured webhook event. |
| events_replay | Replay a captured webhook event to a destination URL (localhost or custom endpoint). |
| endpoints_list | List all webhook endpoints for the authenticated user. |
| endpoints_delete | Delete a webhook endpoint and stop capturing events to it. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.