Solutions

One platform. Eight teams with completely different work.

The same project, the same connectors, the same memory — and eight different jobs underneath. Each team writes the skills for its own work, and nobody stands up a second system to do it.

By role

Start with the work, not the platform.

Every page below is written about that team's actual week — what it can hand off, which connectors make sense for it, and what the output looks like when it comes back.

What never changes

Eight kinds of work. One set of rules underneath.

The pages differ because the work differs. What sits under all of them is identical, and it is worth reading once rather than eight times.

A session is a machine
Every session boots its own disposable Linux computer and cuts its own branch. Thousands can run in parallel on one configuration without touching each other. The machine is real: the agent has a shell, a filesystem and the network.
Work lands through a change request
What an agent means to keep is committed on the session branch and proposed back toward main. Merge is default-deny for agents; an admin can grant project.cr.merge in singulacomp.yaml, and widening that grant is itself a reviewed change.
Approval gates are off until you set them
The shipped default is permissive — an action runs unless you have said otherwise. Set Ask on what should pause and Block on what should never happen, per action or with a pattern rule that can read the arguments in the call.
Connector credentials never enter the machine
The sandbox carries one project-scoped SingulaComp token and no third-party keys. The real credential is decrypted server-side and attached to the outbound call. A runtime secret you deliberately grant is different: it is a real environment value the agent can read.
Everything is a file you own
Agents, skills, connectors, triggers and memory are text in a git repo. You can read the whole company, diff what changed, and revert it. Open source and self-hostable — SingulaComp Cloud, your own VPC, or your own on-prem network.

Run your whole company from one repo you own.

Start with one job and grow from there.

Get started