Your AI Tools Now Know Testkube: August 2026 Release

Most engineering teams have quietly restructured around AI tools. Code gets written in Cursor. Reviews happen with Claude. Copilot fills in the gaps. According to Stack Overflow's latest developer survey, 84 percent of developers now use AI tools in their daily workflow, and real production data shows more than a fifth of merged code is AI-authored.

That changes what engineers expect from every platform they touch. When you ask your AI assistant to set up a tool, write a configuration, or investigate a failure, you expect it to know how. Not to search documentation, guess at syntax, and iterate until something works.

Testing infrastructure has been one of the places where that expectation breaks down. Your assistant can write application code fluently, but ask it to stand up a testing platform, discover the tests already in your repos, or author an execution workflow, and it is operating from general knowledge rather than expertise.

[VIDEO EMBED]

This release closes that gap. Testkube now ships as a set of skills and plugins for the AI tools your team already uses, alongside platform improvements that make Testkube more operable at enterprise scale for both the humans and the agents working in it.

Testing that works where your engineers work

CI/CD tools execute workflows. Testing tools execute tests. Testkube orchestrates testing as a system. And a system is only as useful as it is accessible, which today means being accessible to the AI tools sitting between your engineers and their infrastructure.

Testkube skills, in your AI tool's marketplace

We have published five Testkube skills, open source, that teach AI tools how to work with Testkube correctly: installing the CLI, installing Testkube open source, discovering existing tests in your repositories, authoring test workflows, and running tests and working with executions.

They are available via their GitHub repository at github.com/kubeshop/testkube-skills.

Install the Testkube plugin and your assistant gains real, specific knowledge of how Testkube works. Ask it to discover the tests in a repo and create workflows that run them, and it follows the same patterns our own team uses, rather than reconstructing them from documentation.

The skills are markdown, public, and versioned in the open. As the platform evolves, the skills evolve with it, and your tools pick up the updates automatically. This is the same principle that has always run through Testkube: we do not ask you to adopt our stack. We make Testkube fluent in yours.

Workflow validation built for authors of every kind

Workflow authoring also gets sharper feedback. The workflow editor now returns specific, line-level error messages when a definition is invalid: what is wrong, and where.

For engineers, that means faster authoring. For AI agents creating workflows through the MCP Server, it matters even more. Agents work best with precise, actionable feedback, and line-level validation lets an agent correct a workflow definition immediately instead of iterating toward a valid one. If you are using AI tools to author testing workflows, this release makes them measurably better at it.

Private AI chats

AI interactions in Testkube now follow the same access boundaries as the rest of the platform. When you start an AI chat, you can mark it private, visible only to you. Public chats remain available to the team, and chats created automatically by agent triggers stay shared, since they belong to the event, not a person.

For enterprise teams with many users across many workflows, this keeps AI adoption aligned with how you have structured access in the rest of your environment. Troubleshooting sessions that surface logs and execution detail stay with the people meant to see them.

Insights: from trend to execution in one click

Insights now segments and filters on execution tags. If you tag executions by release, sprint, or failure category, you can build analyses along those lines: failures by category across the last month, aborted runs for a specific release, trends per sprint.

And charts now drill down. Click a segment and you land on the executions behind the number. The distance between "something regressed" and "here is the run that shows it" is now one click, not a hunt through execution lists.

Sharper execution filtering

The execution view adds two new filters: actor type, so you can isolate runs triggered manually, by test triggers, or by other actors, and date range. Combined with the existing label, tag, status, and health filters, you can get to exactly the set of executions you care about, like last week's failed runs from a specific trigger.

Where this goes

The skills we shipped today are the first set, not the last. The pattern is now in place: anything Testkube can do becomes something your AI tools know how to do. Testing has been the stage of software development still waiting on humans. Release by release, we are removing the reasons for that.

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.

The latest enhancements are available now. Sign up and explore the full control plane and look at what your tests have been doing all along.

There is more on the way. Stay tuned.