We’ve released MCP 2.0, a major upgrade that makes agents truly autonomous when working with your Currents data.

Previously, MCP could only check specific run data if you manually provided an ID, meaning you often had to stop and grab info from the dashboard. It also had no access to historical data. With MCP 2.0, that friction is gone. Agents can now discover, analyze, and debug your CI test runs on their own.

What’s New

• Seven new tools for exploring projects CI test runs, specs, and results.

• Smarter workflows like finding flaky or slow tests, without needing run IDs.

• Simplified responses and a cleaner, more modular codebase under the hood.

New Abilities for Agents

With MCP 2.0, you can now ask your agent things like:

• 🔍 “Fix this test” → pull the last CI test results and suggest fixes.

• 🐞 “What were the top flaky tests in the last 30 days?” → analyze CI test history to identify and resolve flakiness.

• ⚡“What were the slowest specs this week?” → find the slowest files across recent CI executions.

• 🧪 “Fix all my flaky tests” → investigate patterns in your CI pipeline, create a plan, and suggest fixes.

In short, MCP 2.0 makes your AI agent behave more like a senior engineer with direct access to your test suite, diagnosing issues, spotting patterns, and guiding you to faster fixes.

We’re just getting started with MCP. Your feedback will help us shape where it goes next, and we’re excited to keep making it smarter, faster, and more helpful with every release.