Testkube AI: More Use Cases, a Smarter UX, and Now available for Open Source

Last month, we shipped Testkube AI. AI-powered Test Creation, AI Agents, and the MCP Server gave engineering teams a new way to create tests, optimize test execution, analyze test results, and connect their favorite AI tools to real test execution data.

That was the starting point.

This release expands on that foundation. We are adding new AI use cases to the dashboard, bringing proper per-user authentication to enterprise MCP workflows, and giving the open source community their first taste of what Testkube AI can do.

One-click AI analysis across executions and workflows

If you have used Testkube AI to analyze a test workflow, you know how it works: open a chat, find the execution or the workflow, figure out the right question to ask. It works, but it assumes you already know what you are looking for.

With this release, there is now an AI Analyze button at both the execution level and the workflow level. Click it, and you get a functional report covering pass/fail trends, resource usage, outliers, and optimization recommendations. No setup, no prompt crafting, no context-switching.

At the execution level, the default agent runs troubleshooting analysis on the selected run. At the workflow level, it analyzes previous executions and gives you a high-level summary of how that workflow has been performing over time: which runs are healthy, where resource usage spikes, and which executions are statistical outliers.

The point is not to replace the deeper AI chat. It is to give you a baseline. Once you see the report, you can start a conversation with the agent from a position of understanding rather than guessing where to begin. From there, you can ask for optimization suggestions, request workflow changes, or dig into specific failures.

We also now surface token consumption in AI chats, so you can see exactly what each analysis costs. Alongside that, this release includes a set of stability and UX improvements across the AI experience that make the whole system more reliable to use day to day.

MCP Server now supports OAuth authentication

The Testkube MCP Server already lets you connect your AI tools (Claude, Cursor, or any MCP-compatible client) to your test execution context. What it did not do, until now, was authenticate those connections properly at the user level.

Previously, connecting the MCP Server required a shared API token. An admin would create the token, distribute it to developers, and manage expiration and rotation manually. Every developer who used that token had the same access level. There was no way to enforce individual permissions or revoke access for a single user without rotating the token for everyone - or managing individual tokens for each developer.

With OAuth support, each developer authenticates through a browser window tied to their Testkube dashboard account. Permissions are enforced automatically based on their role. No shared tokens, no manual distribution, no expiration headaches across large teams.

If you are running the MCP Server across a sizable engineering team, this is a meaningful operational improvement. 

AI analysis comes to Testkube open source

AI analysis is now available in the open source edition. OSS users get one-click AI analysis in the Test Execution Viewer introduced in the previous release. With this release, Testkube OSS provides a structured report covering pass/fail trends, resource usage, outliers, and optimization recommendations. The same baseline that lets you start a conversation with an agent instead of figuring out where to begin.

This matters because openness is not just about the source code. It is about making the capabilities that help teams understand their test health available to everyone running Testkube, regardless of which edition they use. If you are on Testkube OSS today, AI analysis is now part of your experience.

What is coming next

One of the most-requested capabilities from our customers is the ability to track individual test case and performance metrics over time, not just at the workflow level. For example, if you run a workflow with 100 test cases and one of them fails intermittently, you want to see that specific test's pass/fail history, duration trends, and flakiness individually.

We are actively working on this and while some of the foundations for this are already in the product,  we will have more to share about test case-level data granularity and what it enables for your reporting and insights in the next release. Stay tuned!

Get started

We are looking forward to seeing how this release helps your team get more out of Testkube. To explore everything that shipped, check out the full documentation. And in case you missed it, we recently held a webinar where we walked through Testkube AI in detail, covering the full set of capabilities and how teams are putting them to work. You can watch the recording here.

There is more on the way. Stay tuned.

‍