Decisions MCP server

This is a small helper that lets your AI keep a written record of technical decisions. For each decision you can note what you expected to happen, and later write down what actually happened. It is handy for anyone who makes choices about tools or architecture and wants to remember why, and whether the choice worked out.

What this MCP server does

You tell your AI about a decision you made, for example which database you picked and why. The AI passes that to this helper, which saves it in a simple text file on your computer. Later you can ask the AI to add a prediction, like 'this should cut page load time in half'. When you have real results, you tell the AI, and it records whether the prediction was a success, a partial success, or a failure. The helper also reminds you when a decision still has predictions with no outcome yet.

Click to zoom

What you can do with it

• Record a technical decision with the problem, the chosen solution, and the alternatives you skipped
• Attach a measurable prediction to a decision
• Record the real outcome and mark the prediction as success, partial success, or failure
• Search your decisions by keyword, technology, or domain
• See which decisions still have predictions waiting for an outcome
• Get a summary of how a technology has performed across your decisions

Try asking your AI

• “Record a decision: we chose Postgres over MongoDB for the billing service because we need strong transactions.”
• “Add a prediction to that decision: this should reduce duplicate-charge bugs by at least half within three months.”
• “Record the outcome for that prediction: duplicate-charge bugs dropped by about 60 percent, so mark it a success.”
• “Search my decisions for anything mentioning Postgres.”

What it gives back to you

You get short confirmations when something is saved, with an ID like DEC-2026-0001. When you search, you get a list of matching decisions with their details, predictions, and outcomes. If a decision still has predictions with no result, the reply includes a reminder that says how many are waiting. The technology report gives you a simple summary of how each technology has performed so far.

Before you start

Who it's for

Developers, tech leads, and small teams who want a lightweight, honest record of why they chose a tool and whether it worked.