# MCPJam
> Test and debug MCP servers, ChatGPT apps, and MCP apps. Open-source inspector,
> OAuth debugger, evals, and cross-client testing — know your MCP server works
> before your users do.

## Start here
- [How to test an MCP server — the complete guide](https://www.mcpjam.com/mcp-server-testing): MCP server testing validates protocol, tool schemas, agent behavior, OAuth, and cross-client compatibility before production. The full workflow and a checklist.

## Features
- [MCPJam Inspector — test & debug MCP servers locally](https://www.mcpjam.com/features/inspector): Open-source playground for MCP servers: connect a server, run its tools, compare models side by side, read the full waterfall trace. Free and npx-installable.
- [Test Agent Skills + MCP locally](https://www.mcpjam.com/features/skills): Load Anthropic Agent Skills alongside your MCP servers and watch a real model use both. Skills viewer plus an LLM playground, running locally with one npx command.
- [MCP OAuth & EMA debugger](https://www.mcpjam.com/features/oauth-debugger): The MCP OAuth Debugger visualizes every step of the auth flow so you see exactly where it breaks, with spec conformance checks and EMA Cross-App Access tracing.
- [MCP evals — test how models use your server](https://www.mcpjam.com/features/cross-client-testing): Generate an eval suite from your MCP server's tools, run it with a real model, and get accuracy, latency, and diagnosed failure patterns, not only protocol checks.
- [MCP testing SDK — @mcpjam/sdk](https://www.mcpjam.com/features/sdk): Test MCP servers in code with @mcpjam/sdk: connect clients, run unit and end-to-end tests, score agent-in-the-loop evals, and check OAuth conformance.
- [MCPJam CLI — test MCP servers from the terminal](https://www.mcpjam.com/features/cli): Probe, debug, and test MCP servers from the terminal with the MCPJam CLI: run a full health check, OAuth conformance, tool calls, and evals in CI.
- [CI/CD eval gates for MCP servers](https://www.mcpjam.com/features/ci-cd-actions): Gate every PR on how models use your MCP server. Wire MCPJam evals into GitHub Actions, track accuracy over time, and catch behavior regressions early.
- [Chatboxes — human acceptance testing for MCP](https://www.mcpjam.com/features/chatboxes): Hosted, shareable UAT environments that mirror the major AI hosts. Testers break things safely, and their sessions become replayable regression tests.
- [Synthetic acceptance testing with Swarm](https://www.mcpjam.com/features/swarm): Turn loose AI personas that act like your users to acceptance-test your MCP server at scale, with multi-turn journeys across every client, before launch.

## Comparisons
- [MCPJam Inspector vs MCP Inspector — where the overlap ends](https://www.mcpjam.com/compare/mcpjam-vs-mcp-inspector): Honest side-by-side of MCPJam Inspector and the official MCP Inspector: protocol validation, OAuth debugging, tracing, evals, and when to use each. Tested hands-on.
- [MCP evals vs agent evals — what's the difference?](https://www.mcpjam.com/compare/mcp-evals-vs-agent-evals): Agent evals measure the agent you built. MCP evals measure how ChatGPT, Claude, or Cursor use your MCP server — what each covers and when you need both.
- [MCPJam vs Postman for MCP — which should you use?](https://www.mcpjam.com/compare/mcpjam-vs-postman): Honest comparison of MCPJam and Postman for testing MCP servers: protocol inspection, OAuth debugging, model-in-the-loop evals, open source, and when each wins.
- [MCPJam vs mcp-use — build vs test your MCP server](https://www.mcpjam.com/compare/mcpjam-vs-mcp-use): MCPJam vs mcp-use: mcp-use is a framework to build MCP servers and agents; MCPJam is a dedicated tool to test, debug, and eval them. Both open source.
- [MCPJam vs Alpic — test your MCP server vs deploy it](https://www.mcpjam.com/compare/mcpjam-vs-alpic): MCPJam vs Alpic: Alpic deploys, hosts, and monitors MCP servers in production; MCPJam is a deployment-neutral tool to test, debug, and evaluate them.

## Who it's for
- [MCPJam for developers — test MCP servers as you build](https://www.mcpjam.com/for/developers): Reproduce any MCP failure in minutes: run your server through a real agent, read the trace, debug OAuth step by step, and gate PRs with evals. Free and open source.
- [MCPJam for engineering managers — reliability as a gate](https://www.mcpjam.com/for/engineering-managers): Make MCP reliability a merge gate: evals on every PR, failures your team can reproduce, and client coverage maintained for you instead of by your engineers.
- [MCPJam for product managers — launch-ready MCP](https://www.mcpjam.com/for/product-managers): Know if your MCP integration is launch-ready: pass rates on realistic requests, failure patterns named, and tester sessions you can replay before users hit them.
- [MCPJam for AI & platform leads — one MCP quality bar](https://www.mcpjam.com/for/ai-platform-leads): Standardize how every team tests MCP: one platform for local testing, CI gates, and cross-client coverage — open-source inspector plus a hosted team platform.
- [MCPJam for marketers — reach users inside ChatGPT & Claude](https://www.mcpjam.com/for/marketers): ChatGPT and Claude are where your users work now. Ship an MCP integration that performs in every client, with cross-client test evidence before every launch.
- [MCPJam for MCP client teams — parity and coverage](https://www.mcpjam.com/for/mcp-clients): MCPJam tracks what the major AI clients support and how servers behave in each. If you build an MCP client, stay at parity with ChatGPT and Claude.
- [MCPJam for enterprise — SSO, compliance, and support](https://www.mcpjam.com/for/enterprise): Adopt MCP testing without new risk: SSO/SAML, RBAC, audit retention, SOC 2 (in progress), and an SLA. Strictly pre-production — it never touches live traffic.

## MCP clients directory
- [MCP clients directory](https://www.mcpjam.com/clients): Every MCP client we track, with capabilities and testing walkthroughs.

## Latest guides and articles
- [Jeppesen ForeFlight uses MCPJam to iterate 10x faster on AI experiences for pilots](https://www.mcpjam.com/blog/foreflight-mcpjam): How MCPJam supports secure, personalized flight app experiences for the 75% of US pilots using ForeFlight.
- [MCP Inspector V2 Guide: Setup, Debugging & Alternatives](https://www.mcpjam.com/blog/mcp-inspector-v2-guide): MCP Inspector is the official MCP debugging toolkit. See what's new in V2, including Web/CLI/TUI, OAuth, and transports, plus how it compares to MCPJam.
- [Publishing a plugin in the ChatGPT and Codex Plugins Directory](https://www.mcpjam.com/blog/publish-chatgpt-codex-plugin-directory): A plugin allows users to install your MCP server, skills, or both directly in ChatGPT and Codex, without requiring manual configuration.
- [Publishing an MCP server in the Claude Connectors](https://www.mcpjam.com/blog/publish-mcp-server-claude-connectors-directory): There are two ways people can use your MCP server. The first is that you send them a URL and they paste it into Claude's custom connector settings themselves.
- [MCP 2026-07-28 Goes Stateless: What Breaks and How to Migrate to V2](https://www.mcpjam.com/blog/mcp-v2-2026-07-28-migration-guide): MCP v2 goes stateless in 2026-07-28. Learn what breaks and how to migrate sessions, MRTR, Tasks, OAuth, headers, caching, and MCP Apps.
- [Effective MCP, Part 1: The User Value Problem](https://www.mcpjam.com/blog/effective-mcp-part-1): In MCP, your integration can execute perfectly while failing the user completely in practice, yet you have almost no visibility into the actual outcome.
- [MCPJam Inspector Web App](https://www.mcpjam.com/blog/web): MCPJam web app provides easy access to testing your MCP app, ChatGPT app, and MCP server directly in the browser, no downloads required.
- [Share and mock MCP app views](https://www.mcpjam.com/blog/views): Save, share, and mock MCP app views with MCPJam. Track your app's UI progress, collaborate with teammates, and mock data to see your UI in different states.
- [Asana uses MCPJam to build ChatGPT Apps and Claude Connectors](https://www.mcpjam.com/blog/asana-mcpjam): Asana is a work management software helping teams organize their work and track progress across teams. Learn how they use MCPJam to build and test their MCP apps.
- [5 Tips for Building MCP Apps That Work](https://www.mcpjam.com/blog/mcp-apps-tips): Practical lessons for building MCP Apps that feel native across multiple hosts. Learn how to adapt to host environments, control data flow, handle loading states, and more.
- [Test Skills + MCP in MCPJam](https://www.mcpjam.com/blog/skills): Learn how to pair Skills with MCP servers in MCPJam. Test your skills and MCP server together in the LLM playground.
- [Projects are now in MCPJam](https://www.mcpjam.com/blog/projects): Projects are now available in MCPJam, a new way to organize your work in the inspector. Use projects to group your servers and share them with your team.

## Key facts
- The MCPJam Inspector core (client, inspector, CLI, SDK, local evals) is open source and free: https://github.com/MCPJam/inspector
- MCPJam runs locally via `npx @mcpjam/inspector@latest` — no ngrok, no ChatGPT subscription.
- Docs: https://docs.mcpjam.com · App: https://app.mcpjam.com