Bernstein MCP server

Bernstein is a helper that runs other AI coding agents for you, in an organised way. You give it a goal, and it splits the work into tasks, runs them side by side, and keeps a record of what happened. It is handy if you work with AI coding tools and want more control and less mess.

What this MCP server does

You tell your AI what you want done, like fixing a test or building a small feature. Your AI sends that goal to Bernstein. Bernstein breaks the goal into smaller tasks, gives each task to a coding agent, and runs them in separate safe copies of your project so they do not step on each other. It checks the results, records every step, and hands back a summary of what was done and what changed.

Click to zoom

What you can do with it

• Run several coding agents at the same time on different parts of a job
• Keep each agent in its own separate copy of your project so changes do not clash
• Record every step of a run so you can check it later
• Route tasks to different agents based on cost or ability
• Show you a live view of what is running and what is finished
• Produce a signed receipt you can verify offline
• Handle non-code work like reports or datasets with the same tracking

Try asking your AI

• “Fix the failing test in tests/test_foo.py and show me what changed”
• “Run three agents in parallel to add logging to the payment module”
• “Give me a record of everything that happened in the last run”
• “Use the cheaper agent for the simple tasks and the better one for the hard parts”

What it gives back to you

You get back a summary of the run: which tasks were done, which agent did them, and what files changed. If something failed, it tells you where and why. You can also get a signed receipt file that proves the run was not tampered with, which you can check later without needing the original system.

Before you start

Who it's for

Developers and technical teams who already use AI coding agents and want to run them in a more organised, trackable way.