MCP server · Notes

UNITARES MCP server

Give your AI agents one shared record of what they found, who disputed it, and what replaced it after a restart.

Notes For it and development Agreements & Coordination

UNITARES is a program you run on your own computer that keeps a shared notebook for your AI agents. When one agent finds something, disagrees with something, or finishes a task, it writes that into the notebook. Later, another agent (or you) can read the notebook and see where each claim came from. It is handy if you run several AI agents across different tools and they keep losing track of each other's work.

What is an MCP server? The 30-second version

On its own, your AI can only chat with you, and it forgets things between sessions. An MCP server is a small helper program that gives your AI a new skill or a connection to something outside the chat. This server is a helper that connects your AI to your own UNITARES notebook, where findings, reviews, and outcomes are stored. Once it is connected, your AI can look things up in that notebook or write new entries when you ask.

What this MCP server does

You ask your AI to remember a finding, record a result, or look up what another agent said before. The AI passes that request to the UNITARES helper. The helper writes it into your UNITARES record or reads it back from there. When one finding replaces an older one, the helper links them so the older one points to the newer one. You then see a summary in the chat, like the list of findings, the open review notes, or a confirmation that your entry was saved.

What you can do with it

Record a finding with its evidence and sources
Mark an older finding as replaced by a newer one
Search the shared record for what other agents found
Leave a review note or objection on a piece of work
Log a test, build, or task result against the check-in that predicted it
Declare that a new agent continues an earlier agent's task
Ask another connected model for a second opinion and keep the answer
Read open review conditions and outcomes before acting

Try asking your AI

“Save this finding: the upload retry test fails because of clock skew, and link it as a replacement for the earlier flaky test note”
“Search the shared record for anything about upload retry before I start changing the code”
“Record that the build succeeded against the check-in from earlier today”
“Leave a review note on that finding saying the evidence looks thin, and keep it open”

What it gives back to you

You get answers in the chat: a list of matching entries from the record, a short summary of one entry with a link to the entry it replaced, the review notes still open on a piece of work, or a plain confirmation that your write was saved. When you search, you see who wrote each entry and whether it has been superseded. When you record something, the reply usually shows the id of the new entry so you can refer to it later.

Before you start

What you need

Git, curl, and Docker Compose (for the Docker install)
Permission to run a server on your own machine or on a server you control
An MCP client like Claude Code, Codex, or another tool that can connect over HTTP

Good to know

This server keeps a record on your own machine; it does not run your agents or fix their mistakes, and it can store whatever your agents write, so be careful about what they record if the work is sensitive.

Install it with your AI

Add UNITARES MCP server to your AI, no technical skills needed

You don't install anything by hand. You copy one prompt, paste it into an AI that can work on your computer, and it checks, installs and connects the server for you, asking you when it needs something.

Sign in to get the install prompt

Members get a ready-made prompt that lets the Claude desktop app check UNITARES MCP server, install it and connect it for them, step by step. You don't need any technical skills: you copy, paste and answer a few questions. Your connected AI can also find and install any of the 4,609 MCP servers here for you.