Testkube Execution Cost Analyzer Agent

An AI Agent that analyzes execution patterns across your workflows to identify where time and resources are being spent, which workflows are most expensive to run, and where optimization efforts would have the highest impact.

Requirements

An Execution Cost Analyzer Agent requires:

• Access to workflow metrics, execution history, and resource data — provided by the integrated Testkube MCP Server.

This agent uses only the built-in Testkube MCP tools and does not require any external MCP servers.

Optionally, it can also be set up to:

• Run on a schedule via AI Agent Triggers for periodic cost reporting
• Suggest workflow definition changes using update_workflow (with approval)

Create the Execution Cost Analyzer AI Agent

Create an AI Agent as described at Creating an AI Agent, name it what you want and set the prompt to the following (feel free to adapt it to your needs!):

You are a test infrastructure cost optimization expert. Analyze workflow execution patterns to identify inefficiencies and recommend concrete optimizations to reduce execution time and resource consumption.

When asked to analyze costs:

1. Use list_workflows and query_workflows to discover all active workflows
2. Use get_workflow_metrics for each workflow to get execution frequency, duration statistics, and pass rates
3. Use get_workflow_resource_history for the top consumers to analyze resource usage trends
4. Use get_workflow_execution_metrics for detailed CPU/memory time-series data on expensive workflows
5. Use get_workflow_definition to inspect workflow configs and identify optimization opportunities

Analyze and report on:

• Top consumers: Rank workflows by total execution time (frequency x average duration) and resource usage
• Unnecessary runs: Workflows that run frequently but have 100% pass rates (candidates for reduced frequency)
• Over-provisioned resources: Workflows requesting significantly more CPU/memory than they use
• Long-running outliers: Workflows whose duration is disproportionately long compared to similar workflows
• Parallelization opportunities: Sequential workflows that could benefit from parallel step execution
• Redundant executions: Workflows that run on overlapping schedules or duplicate test coverage

For each finding, provide:

• The workflow name and current cost profile (duration x frequency)
• The specific inefficiency identified
• A concrete recommendation with estimated savings
• Implementation complexity (easy, medium, hard)

Present findings as a prioritized table sorted by estimated impact, with a summary of total potential savings.

Enable the following Testkube MCP tools for this agent:

• list_workflows — to discover all workflows
• query_workflows — to search and filter workflows
• get_workflow_metrics — for execution frequency, duration, and health statistics
• get_workflow_resource_history — for resource usage trends across runs
• get_workflow_execution_metrics — for detailed CPU/memory time-series data
• get_workflow_definition — to inspect workflow configuration for optimization opportunities
• query_executions — to analyze execution patterns and frequencies
• list_executions — to examine recent run history
• list_labels — to understand workflow groupings and categories

Using the Execution Cost Analyzer AI Agent

Interactive Analysis

Start a chat with the agent and ask it to analyze costs:

• "Which workflows are consuming the most execution time?"
• "Analyze all workflows labeled team=backend for cost optimization opportunities"
• "What are my top 5 most expensive workflows and how can I reduce their cost?"
• "Find workflows that run more often than necessary"

As a Scheduled Report

Set up an AI Agent Trigger to generate a weekly cost report with a prompt like:

Produce a weekly execution cost report. Analyze all workflows, rank them by total execution time, and highlight the top 10 optimization opportunities with estimated time savings.

Enhancing with External MCP Servers

Connect additional MCP Servers to enrich cost analysis with infrastructure and financial context:

• Kubernetes MCP Server — Query actual cluster resource usage and quotas. The agent can compare Testkube workflow resource consumption against total cluster capacity, identify contention with other workloads, and recommend namespace-level resource adjustments.
• Grafana / Prometheus — Pull infrastructure cost metrics, node utilization, and cloud billing data. This lets the agent estimate actual dollar costs per workflow execution, not just time.
• GitHub / GitLab — Correlate cost increases with recent workflow definition changes. If a PR added a new test step that doubled execution time, the agent can pinpoint exactly when costs increased and who to follow up with.
• Slack — Post the weekly cost report to a #test-infrastructure channel, highlighting the top optimization opportunities with estimated savings.