Put coding agents to work on the machine that has the code.
Hand a job to Claude Code, OpenCode or GitHub Copilot's coding agent on whichever machine you choose. Give each folder its own permission level. Let Sy install the agents and walk you through signing in.
The agents you already use
Sy doesn't replace your coding agent. It runs the one you picked, where the work is.
- Claude Code
- Anthropic's coding agent, installed and signed in for you on any machine you choose.
- OpenCode
- The open-source coding agent, installed and signed in the same way.
- GitHub Copilot's agent
- Hand work to Copilot's coding agent on a chosen machine.
Permissions set folder by folder.
An agent that can read and write everywhere on a machine is an accident waiting for a reason. Syrebrum lets you set a permission level for each folder, so an agent works where you've allowed it and nowhere you haven't.
Shared memory masks secrets, and Sy never handles your passwords, keys or tokens.
Installed and signed in for you
Setting up an agent on a headless server usually means a keyboard, a monitor and a browser you don't have there. Sy does the install and brings the sign-in to you.
Pick the machine
Tell Sy which machine should get OpenCode or Claude Code.
Sy installs it
The agent is installed on that machine for you.
Sign in from your desk
The sign-in link shows up on your main PC, so you finish from where you sit.
Hand it work
Send the agent a job through Sy, like any other task.
Two agents. Two machines. One summary.
Because each agent runs on its own machine, you can give two of them two different jobs at the same time without them fighting over one CPU. Sy brings both results back into the same place.
And because Sy keeps a shared memory of machines, projects and what has been done, the next agent starts with the context the last one left behind. How shared memory works
Questions about coding agents
Do I need my own accounts for these agents?
Yes. Sy walks you through signing in to each agent, and the agent then works under your own account and its provider's terms.
Can an agent run on a machine with no screen?
Yes. Workers can be Ubuntu or Debian servers. When Sy sets up an agent there, the sign-in link appears on your main PC.
Can I limit what an agent can touch?
Yes. Permission levels are set per folder, so an agent only works where you've allowed it.
Does Sy push code or merge anything on its own?
Sy runs the agent you chose, inside the folder permissions you set, and reports back what it did. Read the agent's changes before you ship them, the same way you would without Sy.
Tell Sy what you want done.
YouCheck every machine and tell me what needs fixing first.
Free during the beta. Runs on Windows 10 or 11,
with Ubuntu, Debian and TrueNAS machines as workers.