Currents MCP Server: Turn Your AI into a Test Debugging Partner

Skip the tedious back-and-forth between your CI dashboard and code editor when tests fail. The Currents MCP Server gives your AI direct access to test results, error details, and run metadata so you can ask it to analyze failures and suggest fixes instantly.

The Test Debugging Reality

You know the drill: tests fail in CI, you open the Currents dashboard, dig through logs, copy error messages, switch back to your editor, and try to piece together what went wrong. Then repeat for the next failure. It's death by a thousand context switches.

What This Changes

Instead of playing detective with test failures, you can now ask your AI assistant directly:

• "Why did the login test fail in run abc-123?"
• "Show me all the retry attempts for the checkout spec and suggest a fix"
• "What's causing the flaky behavior in the user-profile tests?"

Your AI gets the complete picture - run metadata, spec file attempts, error details, and retry information - without you having to manually gather and paste data.

Three Tools That Matter

get-run - Pulls complete run information by ID, giving your AI the full context of what happened during test execution.

get-spec-file-attempts-and-errors - Retrieves detailed attempt and error data for specific test instances, perfect for understanding retry patterns and failure modes.

get-api-config - Verifies your API connection is working and shows what endpoints are being used.

Real Scenarios Where This Shines

Debugging Flaky Tests: Instead of manually correlating multiple test runs, ask your AI to analyze retry patterns across failed runs and identify the root cause of intermittent failures.

Batch Failure Analysis: When a deployment breaks multiple test suites, have your AI analyze all the failures at once and group them by likely root causes.

Test Optimization: Get AI suggestions for improving test reliability based on actual failure patterns from your Currents data.

PR Review Context: During code review, quickly get AI analysis of how your changes affected test stability across different environments.

Integration That Actually Works

Works seamlessly with Cursor Editor and Claude Desktop. The setup is deliberately simple - one NPX command and an API key:

{
  "mcpServers": {
    "currents": {
      "command": "npx",
      "args": ["-y", "@currents/mcp"],
      "env": {
        "CURRENTS_API_KEY": "your-api-key"
      }
    }
  }
}

Why This Beats Manual Investigation

Your AI can process and correlate test data faster than you can read it. It can spot patterns across multiple runs, identify common failure modes, and suggest specific code changes - all while you stay in your editor.

The server is lightweight (just three focused endpoints), requires no additional infrastructure, and connects directly to your existing Currents setup. It's built by the Currents team, so it stays current with their API changes.

Get your Currents API key, run the NPX command, and start having actual conversations with your AI about test failures instead of playing data detective.