MCPeek

Get Started
Early Adopter Access

You can now join as an early adopter!

We are currently onboarding users through the waitlist. Join as an early adopter and we will onboard you immediately. You will also get one free month of premium access without creating a subscription.

Join the waitlist to get early access and a guided onboarding from our team.

Join the Waitlist

Examples

Realistic end-to-end examples

Use these preview patterns to wire MCP server instrumentation, ChatGPT widget instrumentation, ingest requests, and dashboard verification.

MCP server integration

server.ts
1import {2  withConnectedMcpServerAnalytics,3  withMcpServerAnalytics,4  withWebStandardMcpServerAnalytics,5} from "@mcpeek/server-sdk";6import { StreamableHTTPServerTransport } from "@modelcontextprotocol/sdk/server/streamableHttp.js";7import { WebStandardStreamableHTTPServerTransport } from "@modelcontextprotocol/sdk/server/webStandardStreamableHttp.js";8 9const apiConfig = {10  endpoint: process.env.MCPEEK_INGEST_ENDPOINT!,

ChatGPT widget integration

todo-widget.html
1// Exposed in the widget bundle (see test/openai-apps/html-app/client-sdk-entry.ts)2window.McPeek.withOpenAiClientAnalytics({3  endpoint: process.env.MCPEEK_INGEST_ENDPOINT,4  apiKey: process.env.MCPEEK_INGEST_KEY,5  organizationId: process.env.MCPEEK_ORG_ID,6  environment: "development",7});8 9// The SDK wraps window.openai methods automatically10const result = await window.openai.callTool("add_todo", {

Verify in dashboard

  1. 1Use the API key and organization ID provisioned for your preview account.
  2. 2Set MCPEEK_INGEST_ENDPOINT, MCPEEK_INGEST_KEY, MCPEEK_ORG_ID in runtime env.
  3. 3Wrap a supported transport or the server connect path with the MCPeek server SDK.
  4. 4Initialize widget with window.McPeek.withOpenAiClientAnalytics(...).
  5. 5Trigger tool calls / widget actions and confirm 202 responses from ingest endpoint.
  6. 6Open dashboard pages (Overview, Tool Calls, Engagement, User Insights) and verify data appears.

Practical recommendation

Start with environment: "development" to see emit logs, then switch to production mode once the full ingest + metrics path is validated.