Building one agent is easy. Running a workforce of them — securely, observably, and at scale — is hard.
Quirq is the environment layer that makes it simple. One click, and your agent gets a runtime, files, tools, memory, an Observatory to see what it’s doing, and a governed perimeter around everything it touches.
Here’s how to get started in under five minutes.
What Quirq is
Quirq is not a model, an agent framework, or a hosted platform. It’s an environment — the place where agent work happens. You bring:
- Your harness — Claude Code, Codex, OpenClaw, Hermes, or any agent framework
- Your model — any model, any provider, self-hosted or API
- Your infrastructure — your laptop, a cloud VM, Kubernetes, or a Terraform-compatible provider
Quirq containerizes it, deploys it in about 60 seconds, and gives you a browser UI to observe everything it does.
Quick start
One-click install
curl -fsSL https://quirq.ai/install | sh
This installs XO Space — the open-source environment layer. Open http://localhost:5002/space/ and you see every project in your workspace, every agent that’s been active, every todo, every cost.
What you get
- Environment — a containerized runtime with files, tools, memory, and an observable record of everything the agent does
- Observatory — a browser UI showing sessions, artifacts, activity, model usage, and cost per project
- Workspace — every folder is a project. Files, git history, todos and live agent activity on one screen
- Multi-agent — run Claude Code, Codex, OpenClaw, Hermes side by side. Switch with one environment variable. See telemetry from all of them regardless of which is active
How it works
Quirq sits between your machine and your agents as an environment layer:
Your machine → Quirq Environment → Agent(s) → Models (any provider)
↓
Observatory UI
The environment provides:
- Runtime — the process, the PATH, the environment variables the agent needs
- Files — workspace access scoped to what the agent is allowed to touch
- Tools — git, shell, MCP servers, APIs, configured per environment
- Memory — session history, project context, todos that survive restarts
- Observability — every action, token, cost, and file touch recorded and visible
Next steps
Once you have it running:
- Point it at a project — run the installer from your workspace directory. Every sub-folder becomes a project
- Configure your agent — set
AGENT_NAME in .env to your preferred harness
- Open the Observatory —
http://localhost:5002/space/ shows everything
- Scale up — add more projects, more agents, connect more models
What people are building
The Quirq ecosystem is growing fast. People are using it for:
- Solo development — one environment for all your coding agents, with cost tracking and session history
- Team workflows — shared environments with governed access and auditable records
- Production deployments — containerized agent fleets on Kubernetes with the Observatory as the control plane
- Experiments — test different models and harnesses side by side, compare costs and outcomes
Video walkthrough
A full Getting Started video is coming soon — covering installation, first environment setup, connecting your first agent, and the Observatory dashboard.
For now, the docs at docs.xo.builders have everything you need to go deeper.
Contribute
Quirq and XO Space are open source. The repos are at github.com/quirq-ai. Issues, PRs, and questions welcome.