Skip to main content
Testkube 2.8.0 is out! Autonomous AI Agents, Custom AI Models, fail-fast and input/output parameters for Workflows, and much more. Read More

Testkube AI Overview

Testkube AI brings intelligent automation and assistance to your continuous testing pipelines. From interactive troubleshooting in the Dashboard to fully automated failure analysis triggered by pipeline events, Testkube AI helps both QA engineers and DevOps teams work faster and more effectively.

AI Assistant

The AI Assistant is your entry point to Testkube AI — an always-available chat interface in the Testkube Dashboard. Open it from the bottom-left button to ask questions, analyze failures, or run any AI Agent interactively.

  • Overlay mode for quick questions on top of your current view
  • Docked mode for side-by-side work during longer analysis sessions
  • Switch between AI Agents and LLM models on the fly

Learn more about the AI Assistant →

AI Agents

AI Agents are the core of Testkube AI — customizable, LLM-powered workers that perform analysis and automation tasks using MCP (Model Context Protocol) tools to interact with your workflows, executions, and external systems.

Default Agents

Every environment comes with four pre-configured agents ready to use:

AgentWhat it does
Testkube HelperGeneral-purpose assistant — ask it anything about your workflows, executions, or environment
TroubleshootAnalyzes failed executions — examines logs, artifacts, and history to identify root causes
Design & OptimizeCreates new Test Workflows and optimizes existing ones (with approval for changes)
Analyze & ReportSummarizes execution trends, workflow health, and metrics into actionable reports

Custom Agents

Build agents tailored to your needs by defining a custom prompt and selecting which MCP tools the agent can access. Connect external MCP Servers (GitHub, Slack, Jira, etc.) to extend agents beyond Testkube.

Learn more about AI Agents →

AI Agent Triggers

AI Agent Triggers turn agents into automated pipeline components that react to events without human intervention. Configure a trigger to automatically run an agent when:

  • A workflow execution fails — instant root cause analysis or automatic rerun
  • A workflow changes state — detect newly broken workflows without repeated noise
  • A scheduled workflow completes — periodic health audits, compliance checks, or cost reports

Triggers support label selectors to scope which workflows activate them, prompt templates with execution context variables, and deduplication to prevent duplicate sessions.

Learn more about AI Agent Triggers →

AI Models

Testkube AI works with your choice of LLM. Platform models are provided out of the box, and you can add your own models (OpenAI, Azure OpenAI, or any OpenAI-compatible service) to use specific models for different agents or tasks.

Learn more about Configuring AI Models →

Testkube MCP Server

The Testkube MCP Server implements the Model Context Protocol standard, enabling AI tools in your IDE — GitHub Copilot, Cursor, Claude, and others — to interact directly with your Testkube environment. Run workflows, analyze failures, and manage tests using natural language from your editor.

Learn more about the MCP Server →


Getting Started

  1. Enable AIConfiguration Quick Start
  2. Try the AI Assistant — Open it from the Dashboard and ask a default agent to analyze a recent execution
  3. Create a custom agentDefine an agent tailored to your workflow, or start from an example
  4. Automate with triggersSet up a trigger to run agents automatically on failures or on a schedule
  5. Connect your IDEConfigure the MCP Server for AI-assisted testing in your editor

Reference