Integrate via MCP

Testzilla exposes a Model Context Protocol (MCP) server so AI agents (Claude Code, Cursor, and other MCP clients) can list projects, create and run tests, and read results on your behalf, all scoped to your account. It is the same backend as the REST API, so anything the API can do is available to an MCP-connected agent.

New to Testzilla?

If you have not run a test yet, start with the no-code Quickstart. This page is for connecting an AI agent to Testzilla, not for first-time onboarding.

Prerequisites

• A Testzilla account with an active subscription. Sign up at app.testzilla.ai.
• An MCP client — Claude Code, Codex, Cursor, VS Code, Lovable, or any other.

The Testzilla MCP endpoint

Every client connects to the same Streamable HTTP endpoint:

https://mcp.testzilla.ai/mcp

Authentication is per-user: you sign in once (OAuth) or present a Testzilla API key, and the server scopes every tool call to your account. See Authentication for how to create an API key.

Add it to your client

Use the tab for your tool. After adding the server, authenticate when prompted, then check the client’s MCP panel to confirm the testzilla_* tools are listed.

Claude Code

Add the server with the CLI:

claude mcp add --transport http testzilla https://mcp.testzilla.ai/mcp

Then run /mcp inside Claude Code to authenticate and confirm the tools are available.

Codex

Codex CLI configures Streamable HTTP servers in ~/.codex/config.toml. Add:

[mcp_servers.testzilla]
url = "https://mcp.testzilla.ai/mcp"

Codex supports OAuth login for Streamable HTTP servers. To use an API key instead, point bearer_token_env_var at an environment variable holding your key. Run /mcp in the Codex TUI to see active servers.

Cursor

Add the server to .cursor/mcp.json (project) or your global Cursor MCP config:

{
  "mcpServers": {
    "testzilla": {
      "url": "https://mcp.testzilla.ai/mcp"
    }
  }
}

Reload Cursor; it will prompt you to authorize (OAuth) on first use.

VS Code

Add the server to .vscode/mcp.json (workspace) or your user mcp.json:

{
  "servers": {
    "testzilla": {
      "type": "http",
      "url": "https://mcp.testzilla.ai/mcp"
    }
  }
}

Or from the terminal:

code --add-mcp "{\"name\":\"testzilla\",\"type\":\"http\",\"url\":\"https://mcp.testzilla.ai/mcp\"}"

Lovable

In Lovable, open Connectors, go to Chat connectors, and click New MCP server. Enter:

• Server Name: Testzilla
• Server URL: https://mcp.testzilla.ai/mcp

Lovable uses OAuth by default — click Add & authorize and complete the sign-in. (You can also authenticate with a bearer token / API key if you prefer.) The connector then appears in your chat connectors list.

Other clients

Any MCP client that supports a remote Streamable HTTP server works. Give it the endpoint:

https://mcp.testzilla.ai/mcp

Choose OAuth, or supply your Testzilla API key as a bearer token, per the client’s instructions.

Run your first commands

Once connected, ask the agent in plain language, for example:

• “list my projects”
• “run the smoke test on my support line”
• “show me the result of the last run”

What the agent can do

The server scopes every action to your account. The available tools cover:

• User: testzilla_get_user_info, testzilla_get_balance
• Projects: testzilla_list_projects, testzilla_get_project, testzilla_create_project, testzilla_update_project, testzilla_set_project_prompt, testzilla_delete_project
• Channels: testzilla_list_channels, testzilla_get_channel, testzilla_create_channel, testzilla_update_channel, testzilla_set_channel_prompt, testzilla_delete_channel (create/update accept the adapter-transport fields: chat_url, agent_id, vapi_public_key/vapi_assistant_id, elevenlabs_agent_id, widget selectors, screenshot options, target_channels, etc.)
• Suite Runner: testzilla_run_folder — run a Folder’s (adapter none) test suite across target channels as a matrix
• AI reports: testzilla_generate_channel_report, testzilla_generate_project_report
• Tests: testzilla_list_tests, testzilla_get_test, testzilla_create_test, testzilla_update_test, testzilla_delete_test, testzilla_run_test
• Queue: testzilla_list_queue, testzilla_get_queue_item
• Results: testzilla_get_test_results, testzilla_get_result
• Batches: testzilla_list_batches, testzilla_get_batch, testzilla_create_batch, testzilla_update_batch, testzilla_delete_batch, testzilla_run_batch, testzilla_get_batch_runs, testzilla_get_batch_run
• Console: testzilla_console_create_session, testzilla_console_execute, testzilla_console_quick, and related session tools
• AI analysis: testzilla_analyze_failures, testzilla_analyze_test
• Git integration: connect a repo and sync test prompts (the testzilla_git_* tools)

Note

The MCP server is the same backend as the REST API, so anything you can do with the API is available to an MCP-connected agent, scoped to your account.

No client? Use the Console instead

You do not need to wire up an external MCP client to get the agent experience. The built-in Testzilla Console (Tessie) runs the same tool set from inside the app — describe what you want in plain language and Tessie creates projects, channels, tests, and runs, reads your connected systems through Integrations, manages schedules, and improves your prompts from real results. The testzilla_console_* MCP tools expose that same console to external agents, so an MCP client can drive a Console session programmatically.

Use the Console when you want the no-code power surface in the browser; use an MCP client (above) when you want Testzilla available inside your own coding agent.

Next steps

API Reference

Every endpoint behind the MCP tools, with a built-in playground.

Integrate via API

Automate Testzilla from CI over REST.

Authentication

API keys and tokens explained.