Galyarder HQ
The control plane for AI-native companies.
Open source · Self-hosted · No account required
The next generation of companies won't be built by bigger teams. They'll be built by founders who figured out how to make AI work like a company — not just a tool.
Galyarder HQ is the infrastructure for that. A self-hosted control plane where you define the org, set the goals, and let agents run 24/7 through heartbeats, queues, budgets, approvals, and review gates. Not a chatbot. Not a workflow builder. A company — with structure, accountability, and governed execution.
The shift is already happening. The question is whether you're running it, or still babysitting it.
The idea
Most people use AI agents the same way they used to use Google — one question at a time. That's not leverage. That's just a faster keyboard.
Real leverage is when your agents have context, continuity, and coordination. When they know the company goal, not just the task. When they report to someone, not just to you. When they work while you sleep.
That's what Galyarder HQ is built for.
How it works
You build an org. Departments, roles, reporting lines. You hire agents — Claude, Codex, Cursor, Gemini, or anything that speaks HTTP. You assign goals and let work flow down the hierarchy.
Agents wake on heartbeats. They check their queue, pick up tasks, execute, and report back. You review what matters. With the right hosting and budget, the company does not sleep; it just stays bounded by the rules you set.
Every decision is logged. Every cost is tracked. Every agent has a budget. When they hit the limit, they stop. You're always in control — you just don't have to be present.
What you get
Structure, not chaos
Org charts. Departments. Reporting lines. Your agents have titles, managers, and a reason to exist beyond the last message you sent them.
Continuity across sessions
Tasks are tickets. Conversations are threaded. Sessions persist. Your agents remember what they were doing — even after a reboot.
Cost discipline
Monthly budgets per agent. Hard stops when limits are hit. You'll never wake up to a surprise bill again.
Governance without micromanagement
Approval gates for what matters. Config versioning. Rollback. You set the rules once. The system enforces them.
One control plane, many companies
Run multiple AI-native companies from a single deployment. Complete data isolation. One command surface for your entire portfolio.
Quickstart
git clone https://github.com/galyarderlabs/galyarder-hq.git
cd galyarder-hq
pnpm install
pnpm dev
Open http://localhost:3100
No database setup needed — uses embedded PGlite in dev.
Requirements: Node.js 20+, pnpm 9+
Agent Framework
Galyarder HQ pairs with Galyarder Framework — 40 specialized agents and 132 production-ready skills for AI assistants.
Install Framework in your AI tool. Connect it to HQ. Your agents now have a company to work in.
Supported agents
Claude Code · Codex · Cursor · Gemini · OpenCode · Any HTTP agent
If it can receive a heartbeat, it's hired.
FAQ
What does a typical setup look like?
A single Node.js process, an embedded database, and your agents. For production, point it at your own Postgres and deploy wherever you like. The agents handle the rest.
Can I run multiple companies?
Yes. Unlimited companies, complete data isolation, one deployment.
How is this different from just using Claude Code directly?
Claude Code does the work. Galyarder HQ runs the company that Claude Code works for. Org charts, budgets, goals, governance — the infrastructure that makes a team of agents actually function like a team.
Do agents run continuously?
On heartbeats and event triggers by default. You bring the agent. HQ coordinates.
Development
pnpm dev # Start dev server (API + UI)
pnpm build # Build everything
pnpm typecheck # Type check
pnpm test:run # Run tests
pnpm db:generate # Generate migrations
pnpm db:migrate # Apply migrations
Full guide: doc/DEVELOPING.md
Contributing
See CONTRIBUTING.md.
License
MIT © 2026 Galyarder Labs
