• LOGS
  • VISUAL
  • SUMMARY
  • HUMAN

How to debug failed test executions with your AI

Connect your agent to Perfecto once. After that, you just ask your AI to find out why a test failed.

Ask AI about this page

Short answer

How can I debug failed tests using my AI?

Your agent pulls error logs and screenshots from Perfecto to show you exactly what went wrong. You get the root cause without having to navigate the web dashboard. This breakdown shows you how it works.

Debugging outcomes

Where your data lands.

The AI organizes Perfecto data into these four outcomes.

LOGS

Error analysis

The agent grabs the specific failure logs. You get the stack trace or error message needed to fix the code.

VISUAL

Visual evidence

Your AI pulls screenshots from the execution. You see the application state at the moment of failure.

SUMMARY

Run overview

The agent reads the execution summary. You get a quick count of passes and failures for the suite.

HUMAN

Manual triage

The AI flags issues that need a person. You decide if a failure is a real bug or just a flaky test.

The workflow

What your AI does when a test fails.

The agent handles the data retrieval so you can focus on the fix.

  1. Fetch execution data

    The agent looks up the specific details of a failed run to understand the context.

    get_execution_details

  2. Gather visual evidence

    The agent finds and retrieves screenshots or video files attached to the run.

    list_artifacts

  3. Review test results

    The agent pulls the summary report to see the overall health of the test suite.

    get_report_summary

  4. Identify the root cause

    The agent combines logs and screenshots to explain why the test failed.

    get_execution_detailslist_artifacts

Try it

Copy these to start.

Use these prompts to begin debugging immediately.

Starting points

Swap the bracketed text for your actual Perfecto execution IDs or test names.

Perfecto Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_qkMLVDlhWeRHbw49znraw9alYWakOJ9o2dj1pqaq/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — Perfecto capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "perfecto-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_qkMLVDlhWeRHbw49znraw9alYWakOJ9o2dj1pqaq/mcp"
    }
  }
}
Copy into chat04

  • Show me the failure logs for execution ID 55432.

  • Get the screenshots for the last failed test in the 'Checkout Flow' suite.

  • Summarize the results for the most recent Perfecto test run.

  • Look at the artifacts for execution 8821 and tell me why the button click failed.

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect Perfecto once, then ask.

You only need to link your account once. Your credentials stay encrypted, and you can start asking questions in your preferred AI client immediately.

Connect Perfecto to your AI

FAQ

How this task behaves

  • 01

    Can my AI fix the code for me?

    No. The AI only pulls logs and screenshots from Perfecto to help you understand the failure. You must perform the actual code changes.

  • 02

    Does the AI delete any test results in Perfecto?

    No. The AI only has read access to execution details, artifacts, and summaries. It cannot modify or delete anything in your Perfecto account.

  • 03

    Can I see the screenshots directly in my AI chat?

    Yes. When the agent uses the tool to list artifacts, it can provide the links or data needed to view the visual evidence of a failure.

  • 04

    Can the AI run new tests in Perfecto?

    No. This specific setup is for reading existing execution data. It cannot trigger or start new test runs.

  • 05

    How does the AI know which test failed?

    You tell the AI which execution ID or test suite to look at, and it uses the available tools to fetch that specific data.

  • More questions about Perfecto? See everything the Perfecto Connector can do

Connect Perfecto to Claude, Cursor, ChatGPT & more