Issues straight to agents
Hand a GitHub, Jira or Linear issue to an agent in one click. The PR comes back to you.
Coding agents that work 24/7.On your own server.
cezar runs Claude Code, Codex, OpenCode and other coding agents in parallel on your laptop or a VPS, on the subscriptions you already pay for. No API keys. Queue the backlog, close the laptop, review the PRs from your phone.
macOS, Linux and WSL2, Node 20+. Free and open source (MIT). No account, no database. Watch the demo
Introducing cezar
A coding team that never logs off.
Queue tasks, hand over GitHub, Jira or Linear issues, or let automations start runs on a schedule. cezar drains the queue in parallel worktrees, runs your checks after every step, and has the PRs waiting for review when you are back.
The night shift
Close the laptop. The work keeps moving.
Schedules, GitHub polls and tracker events start runs while you are offline. When every slot is busy, new runs wait in the queue. Every run gets its own worktree and its checks, and the PRs wait for you in the morning.
- 19:10: GitHub issue #128 started a run on Claude Code.
- 21:40: Claude Code: checks passed. Draft PR #132 opened.
- 22:30: Dependency bump started a run on Codex.
- 23:40: Flaky tests started a run on OpenCode.
- 23:50: Codex: npm test failed. The error goes back to the agent.
- 01:10: Linear ENG-42 is queued #1. Both slots are busy.
- 01:40: Codex: checks passed. Draft PR #133 opened.
- 01:50: A slot is free. Linear ENG-42 starts on Claude Code.
- 03:00: OpenCode: checks passed. Draft PR #134 opened.
- 04:20: Claude Code: checks passed. Draft PR #135 opened.
- Good morning. 4 PRs are ready for your review. Nothing merged without you.
Command it from your phone
Start tasks, answer questions and review the diff from wherever you are.
Dispatch
Think using big model. Spend using small one.
Switch on Dispatch and a big task splits itself. The thinking stays on a strong model, the routine work goes to a cheaper one, and every result reports back to you.
Costs are illustrative.
Take the tour
Click through cezar.
Describe it
Type what you want done, in plain words.
The problem
Your agents stop when you do?
Laptop-bound
The agent sleeps when your laptop does
Agents live in a terminal session. Close the laptop, lose Wi-Fi or go to bed, and the run dies with it.
One at a time
One terminal, one task
The rest of the backlog waits in your head, not in a queue. Nothing moves unless you are typing.
No guardrails
Done means it said done
Nothing runs your tests before the agent calls it finished. You find the breakage in review.
Token bills
Automation costs extra
Orchestrators usually want API keys billed per token, on top of the subscription you already pay for.
What is cezar
One cockpit for all your coding agents. Queue the work and walk away.
Config steps
0
No setup. cezar picks up the agent CLIs you already have installed
Uptime
24/7
One command puts it on a VPS with HTTPS and a login
Agents
Mix
Claude Code, Codex, OpenCode and more. Pick one per step
Price
$0
MIT open source. You pay only for the plans you have
FAQ
Frequently asked questions.
What is cezar?
An open-source orchestrator for AI coding agents. It runs Claude Code, Codex, OpenCode and other coding agents in parallel, each task in its own git worktree, on your laptop or 24/7 on a server, with a web cockpit to watch, steer and review everything.
Do I need API keys?
No. cezar drives the agent CLIs you are already logged into, so runs use the subscriptions you already have. There is no cezar account and no database either.
Which agents can it run?
Claude Code, Codex, OpenCode and pi today, with more added as new agent CLIs come out. Pick one per task or per workflow step, and switch a task’s runner, model or account mid-thread without losing the conversation.
How does it keep working when my laptop is closed?
Run npx cezar-run server-install --platform ubuntu-vps. It sets up HTTPS, a login and a system service, so the cockpit is reachable from anywhere, phone included. Prefer not to run a server? Try it on Open Mercato Cloud.
How many agents can run at once?
As many as your machine can handle. Set a parallel limit for the workspace or per project and the rest waits in a queue that survives restarts. Running tasks can dispatch subtasks of their own, and a memory ceiling pauses heavy runs so the queue keeps moving.
Will it ship broken code?
Not without you. Workflow checks like npm test gate every step (exit 0 means pass) and failures go back to the agent to retry. Every run with changes waits at a review gate, and nothing merges on its own.
What can start a run without me?
Automations: a schedule, a GitHub poll, a PR review, or a Jira or Linear event. With Autonomous on, the run never stops to ask and finishes with the PR ready for review.
What does it cost?
cezar is free, MIT-licensed open source. You pay only for the agent plans you already have, and the Dashboard shows reported usage and cost per task and project.
Who builds it?
cezar was created by Patryk Lewczuk at Open Mercato, the open source CRM/ERP framework company founded by Piotr Karwatka, who previously co-created Divante, Open Loyalty and Vue Storefront (later Alokai, a Y Combinator company with a $40M Series A). Development happens in the open at github.com/open-mercato/cezar.
Can I try it without logging in to an agent?
Yes. CEZ_DRY_RUN=1 npx cezar-run starts the cockpit with a built-in mock agent, so you can look around first.
Put your agentson the night shift.
Run one command and the agents you already pay for get a task queue, workflows that run your tests, and a cockpit you can open from anywhere.
You need Node 20+ and at least one logged-in agent CLI, such as claude, codex or opencode.