MCP Server 3.0: ChatGPT, Claude & Sharing

Testomat.io MCP Server 3.0 can connect to ChatGPT and Claude on the web directly to your testing projects, without requiring a local installation. A smaller, flexible toolset reduces token usage and preserves more context for AI-assisted workflows, while more stable authorization, cross-project test sharing, and richer analytics make it easier to manage, reuse, and understand your testing data.

Remote Access for AI Assistants

Connect web-based AI assistants to Testomat.io using a hosted Streamable HTTP endpoint. All they need is the project-specific URL—there is no MCP package to install and no local server to maintain.

OAuth 2.1 handles authorization for connectors that cannot provide a static project token, opening Testomat.io projects to browser-based assistants such as claude.ai. IDE, command-line, and CI integrations can continue using project tokens, while self-hosted installations retain the existing stdio transport.

More Stable Authorization Sessions

MCP Server 3.0 requests reauthorization only for invalid or revoked tokens. Other access errors—such as restricted permissions or unavailable subscription features—are returned directly to the assistant, preventing unnecessary reconnection loops and clarifying authorization failures..

Cross-Project Test and Suite Sharing

Let AI assistants share tests and complete suites between Testomat.io projects. Content can be selected by IDs or labels, and suites can be distributed to multiple target projects, making it easier to reuse centrally maintained test coverage.

Linked copies remain read-only and synchronized with their source, helping teams avoid duplicates and inconsistent updates. When a shared test or suite needs to evolve independently, it can be unshared and converted into a regular editable copy without changing the original.

A Smaller, More Efficient Toolset

Work with one MCP tool per Testomat.io entity instead of a separate tool for every operation. Actions such as listing, creating, updating, sharing, or deleting are now expressed through a consistent command parameter.

This reduces the interface from 94 operation-specific tools to 17 entity tools. Sending fewer tool definitions leaves more of the AI model’s context available for the task, while a smaller and more predictable interface reduces the chance of selecting the wrong operation during longer sessions.

MCP Server 3.0 changes the tool-calling interface. Existing prompts or automations using names such as tests_list must switch to the tests tool with command: "list". The release also restores reliable attachment uploads and makes tag search properly available to connected assistants.

Configurable Tool Profiles

Control how much of Testomat.io an AI session can access by choosing a full, core, or read tool profile. The consolidated entity-based structure works with these existing profiles, so teams can expose the complete MCP interface, focus on the most commonly used entities, or provide read-only access.

A smaller profile further reduces the tool definitions sent to the AI model and helps keep token-sensitive sessions focused. Read-only mode is also useful when an assistant should analyze project data without being allowed to create, update, or delete it.

Run Insights and Enterprise Analytics

Understand where failures are concentrated by breaking down a test run by suite, tag, label, assignee, or priority. This helps AI assistants identify the parts of a release that need attention instead of returning only overall pass and fail totals.

Enterprise analytics now supports grouped reports for flaky, slow, failing, skipped, and never-executed tests. New milestone reports cover completion, tests, runs, plans, requirements, and contributors, while expanded priority reports provide additional context for release and sprint analysis.

Test run results can also be sorted to surface the latest executions first, and pagination makes larger analytics reports easier to explore. Enterprise analytics remains available through the @testomatio/mcp-enterprise package and requires the corresponding analytics subscription feature.

Find more details at https://testomat.ai/mcp.
Technical documentation - here.

🚀 Create a demo project for free and check all amazing features right now. We look forward to hearing what you think, and what other features we should do!