Skip to main content

The Testmo MCP

The official Testmo MCP connects your Testmo test repository to any AI assistant that supports the Model Context Protocol (MCP) — like Claude Code, Claude Desktop, Cursor, VS Code, or Codex. Once connected, your AI assistant can read from and (with the right access) write to Testmo using natural language, turning your test repository into a live knowledge base for QA and for AI-assisted coding.

This article is the starting point for teams in the MCP closed beta. It covers what the MCP can do, how to connect it, and where to find the tools and workflows it exposes.

This is a closed beta. The Testmo MCP is available to selected teams while we refine it toward general availability. If you'd like access, fill out the form here: Testmo MCP Beta

What you can do with it

Once authenticated, the Testmo MCP exposes a set of tools built on Testmo's existing API, so your AI assistant can work with projects, milestones, runs, results, sessions, automation, and your case repository. Typical workflows include:

  • Analyze existing test coverage for a new requirement and identify the gaps
  • Use your tests as context so an AI coding assistant writes code that respects existing behaviour
  • Generate and organise new test cases in the repository from natural-language descriptions
  • Create runs, record results, and manage exploratory sessions programmatically

The complete, always-current list of available tools is published on the Testmo MCP page. Because the tool list is generated directly from the API surface exposed to the MCP, it stays in sync automatically — there's no separate per-tool documentation to fall out of date:

Read-only vs. read-write access

When you connect, you choose the level of access the MCP has to your Testmo instance:

  • Read-only — the assistant can search and read your repository (cases, folders, runs, results, sessions) but cannot change anything. Best for analysis, coverage review, and using tests as coding context.
  • Read-write — the assistant can also create and update Testmo data (for example, create test cases or record results). Required for workflows that write back to Testmo, such as importing cases.

Access is enforced by the tools the MCP exposes for your connection. Grant read-write only when you intend to let the assistant modify Testmo data.

Getting started

1. Connect the MCP to your assistant

The Testmo MCP uses OAuth for authentication. When you first connect, you'll be redirected to your browser to sign in to your Testmo instance and approve access; your assistant then receives a secure token it presents with each request. Tokens can be revoked at any time from your Testmo Admin settings.

For Claude Code, add the beta MCP endpoint:

claude mcp add testmo "https://testmo.sembi.com/mcp" --transport http --scope user

Other MCP-compatible assistants (Cursor, VS Code, Codex, Claude Desktop) connect to the same endpoint through their own MCP configuration. Refer to the guide here for more info: Testmo MCP page

2. Install the Testmo skills (optional)

Beyond the standard tools, Testmo ships a set of packaged skills — guided workflows that combine several tools to accomplish a specific QA-plus-coding task. To install them in Claude Code:

/plugin marketplace add SembiIQ/sembi-iq-plug
/plugin install testmo@sembi-iq
/reload-plugins

Tools vs. skills

It's worth understanding the difference:

  • Tools are individual capabilities backed by the Testmo API — creating a run, reading a folder, searching cases, and so on. Your assistant calls them as needed. The full list lives on the MCP page.
  • Skills are opinionated, multi-step workflows built on top of those tools. Each has a clear intent and a recommended way to run it, and is documented separately (below).

Featured agent tools & skills

Most tools are self-describing from the tool list. A few capabilities work as agents — they carry their own intent and are worth reading about before you use them:

Security & privacy

How Your Data Is Used

We are committed to building AI features that are secure, ethical, and user-centric. When you use the Testmo MCP Server:

  • Your input—along with any optional context you provide (such as domain terms or project-specific information) is securely transmitted via encrypted APIs to process your request.
  • That input is used solely to generate relevant responses for your request. No data is selected or processed automatically. Only what you submit.
  • Using the MCP Server is entirely optional, and you decide what data is shared.

Review our documentation below for more information:

Was this article helpful?