Testkube AI Agents
Testkube allows you to define AI Agents to perform advanced test orchestration and analysis tasks, using both the internal and external MCP Servers for whatever scenario you need.
Example Use Cases for AI Agents
Example Use Cases for AI Agents are:
- Automated Remediation: introspection of failed tests and related code changes to suggest fixes (see Remediation Agent Example)
- Anomaly Detection: analysis of execution results or resource usage over time to detect anomalies.
- Advanced Integrations: integrating with external systems via their MCP Server(s) for creating issues, sharing/reading data, etc.
- Advanced Troubleshooting: including externally available data for troubleshooting failed tests (see Flakiness Analysis Agent Example)
- Automated Test Generation: Agents can both create new automated tests and the Workflows required to run them.
The only limit to an AI Agents capabilities is set by the LLM you use and the MCP Servers you integrate with.
Testkube AI Agents use the same LLM that has been configured for the AI Assistant for resolving the provided prompt, see Default LLM and Model.
Types of AI Agents
Testkube AI Agents can be designed with two distinct approaches depending on your use case:
Task-Focused AI Agents
Task-focused agents are designed to perform a specific task given specific inputs, typically as part of an automated pipeline. These agents:
- Receive well-defined inputs (e.g., a workflow execution ID, test results, or error logs)
- Perform a specific analysis or action
- Produce a concrete output (e.g., post a summary to Slack, create a GitHub issue, update a dashboard)
Examples: An agent that analyzes a failed workflow execution, identifies the root cause, and automatically posts a summary with remediation suggestions to a Slack channel.
Task-focused agents are ideal for automation scenarios where you want to augment your Continuous Testing pipelines with AI-powered analysis and actions without human intervention.
See the Remediation Agent Example for an example of a task-focused agent.
Guidance and Support AI Agents
Guidance agents are designed to provide interactive assistance for general tasks, acting more like an AI-powered assistant or copilot. These agents:
- Help users accomplish broader goals through conversation
- Provide recommendations, best practices, and step-by-step guidance
- Can perform actions on behalf of the user when requested
Example: An agent that helps users create new TestWorkflows by asking clarifying questions about their testing requirements, suggesting appropriate configurations, and generating the workflow definition.
Guidance agents are ideal for onboarding, exploration, and complex configuration tasks where users benefit from interactive support and contextual recommendations.
Predefined Guidance AI Agents
Testkube includes a set of pre-defined Guidance AI Agents in the Dashboard that you can easily add to your environment and use to get started with AI Agents - read more at Creating an AI Agent from a Template.
Choosing the Right Approach
| Aspect | Task-Focused | Guidance & Support |
|---|---|---|
| Trigger | Automated (e.g., on execution failure) | User-initiated conversation |
| Input | Specific, well-defined data | Open-ended questions or goals |
| Output | Concrete action or artifact | Recommendations and assistance |
| Interaction | Fire-and-forget | Interactive, multi-turn |
| Best for | CI/CD/GitOps automation, alerting | Onboarding, configuration, exploration |
Both styles can leverage the same MCP Servers and tools - the difference lies in how the agent is triggered and how it interacts with users or systems.
Check out Defining AI Agents for more details on how to create and manage AI Agents.