Introducing Syvon Brain.

Context.

One shared intelligence for every agent you run.

Free and open source. The workspace behind Surface, Studio and MCP.

The Workspace Brain

One workspace. One brain.

A Syvon workspace is everything your agents know and can do, kept as files: the brand, the material, the method, the tools and the work. Every agent you plug in reads the same brain, so nothing starts from a blank prompt.

The Brain, in Surface

Every folder, one hub.

This is the Brain screen from Surface. Pick a folder to see what it holds, who writes it and who reads it.

BrainSyvon
SyvonWorkspace brain

Set

The folders

What each part of the brain does.

Each folder is a junction: something writes into it, something reads out of it.

The folder tree

Plain folders, on disk.

The same brain, as Studio shows it: a folder tree you can open and close. Pick a folder to see what it is for.

Content

content

What the work says.

The raw, purely semantic material every design draws from: the copy, the references, the variables. No design markup, so one truth can feed a post, a deck and a reel without being rewritten inside each.

Questions

FAQ

It is your workspace, seen as one thing: the brand, the material, the method, the tools and the work, kept as files every agent can read. Surface draws it as the Brain screen. Studio opens it as folders. Claude reaches it over MCP.

Yes. Its core is free and open source under the MIT licence: the workspace structure, brand DNA, agent permissions, approvals, revisions and feedback signals. The code is on GitHub at github.com/Syvon-ai/syvon-brain. The Brain screen shown on this page is part of Surface.

In files. On your machine when you work in Studio, synced to the cloud so Surface, your team and every render read the same copy.

Syvon’s own agent in Surface, your coding agent in Studio (Claude Code, Codex or OpenCode), and Claude or ChatGPT over MCP. They all read the same brain.

No. Drop material in and the agent files it: uploads land in assets/, brand setup writes config/ and meta/, and what a run learns goes to knowledge/ on its own.

Knowledge is what the workspace has learned: rules written by the run that learned them. It is kept out of ordinary browsing so it stays an honest record, but the Brain shows it.

Syvon Brain

Give Your Agents a Brain.

The Brain core is free and open source. Read the code, then connect your agents to it over MCP.

Connect Brain