cezar

Compatibility note: GitHub list rows may omit body in the next minor release. Consumers should hydrate full text from /api/v1/github/comments/:kind/:number (detail.body) now.

Features

• 💯 Free and open source.
• 🖥️ Uses your own claude, codex, copilot, cursor, agy (Antigravity), gemini, junie, kimi (Kimi Code), omp, opencode or pi login. No API key needed.
• ☁️ Easy to set up on a VPS, so your agents keep working when your laptop is closed.
• 📱 Fully responsive. Start and review tasks from your phone.
• 🔀 Every task gets its own git worktree, so several agents can work at the same time. Extra tasks wait in a queue.
• 🤖 Turn on Autonomous and a run never stops to ask. It just finishes.
• 📡 Watch it work live: agent text, tool calls, tokens and cost.
• 🏁 Run the same task ×2 or ×3, compare the diffs and keep the best one.
• 🧩 Skills are Markdown files and workflows are short YAML files. Mix agents per step.
• 🐙 Run the agent straight on a GitHub issue. Nothing merges on its own.
• 📂 One cockpit for all your projects.
• 📊 A workspace dashboard shows what needs you, what is running and what finished across your projects.
• 💸 Track reported cost and token usage by project in Usage & cost.
• ⏰ Schedule recurring work or launch tasks from GitHub and supported tracker events with Automations.
• 🌳 Agents can delegate independent work to child tasks, each in its own worktree, and receive their reports in the parent session.
• 🎫 Connect Jira or Linear to browse issues and launch tasks from your project tracker.
• 💾 No database. Everything is saved as plain files in .ai/cezar/.

Screenshots

Parallel tasks — Run and queue many tasks, each in its own git worktree.

Live run — Every step, tool call and token, as it happens.

Variants — Run a task ×2 or ×3 and keep the best diff.

Workflows — Build a task's pipeline as a node graph: agents, checks, loops, gates, PRs and CI, saved as YAML.

GitHub — Hand an open issue to the agent in one click.

Skills + Autonomous — Pick a playbook, flip Autonomous and walk away.

Dashboard — See what needs your input or review, what is running and what finished across your workspace.

Usage & cost — Compare reported spend and token usage across projects.

Automations — Schedule maintenance and reviews, and see the week ahead alongside GitHub triggers.

On your phone — the same cockpit, from the task list to the diff.

Quick start

You need Node 20+ and at least one agent CLI you're logged into: Claude Code, Codex, GitHub Copilot CLI, OpenCode, Cursor Agent, Google Antigravity CLI, Junie, Kimi Code, OMP, pi or Gemini CLI (needs an API key, Vertex AI or a Workspace licence — personal Google sign-in no longer works for it). git and gh are optional.

cd your-repo
npx cezar-run

This opens the cockpit at http://localhost:4321. If a compatible cockpit is already running, the command reuses it, adds the current project when needed, and opens that project's page instead of starting a duplicate server. Type a task, pick a workflow, then hit Start. npx cezar-cli still works too — it's the same package under its older name.

npx cezar-run run "add a --json flag to the export command"   # headless, no browser
npx cezar-run init                                            # scaffold .ai/cezar/
npx cezar-run@nightly                                         # try tonight's build

Just want to look around? Run CEZ_DRY_RUN=1 npx cezar-run. It uses a built-in mock agent, so you don't need to log in.

Run it on a server

npx cezar-run server-install --platform ubuntu-vps

This sets up HTTPS, a login and a system service, so you can open the cockpit from anywhere, including your phone. There are guides for Ubuntu VPS and macOS + ngrok.

How it works

1. You describe a task. Type it, attach files, or start from a GitHub issue.
2. cezar runs a workflow (agent steps plus shell checks) in a new git worktree, using your agent CLI.
3. The cockpit streams every step live. If a check fails, the agent tries again and sees the error.
4. You check the result. Read the diff, send notes back, or open a draft PR.

A workflow is a small YAML file in .ai/cezar/workflows/:

name: fix-and-verify
steps:
  - id: implement
    prompt: "{{task}}"
    skill: project-conventions   # optional: a Markdown skill from .ai/skills
    runner: codex                # optional: which agent runs this step
  - id: verify
    command: "npm test"          # exit 0 = pass
    onFail: { retry: implement, max: 2 }

The built-in quick-task workflow runs with no setup. A check can also be a browser: browser and mobile e2e as a verification step turns an agentic e2e run — or an independent QA exploration — into the gate a task has to pass.

Automations

Turn repeatable work into an automation: check dependencies every morning, draft release notes on Fridays, or review each new pull request. Each match or scheduled occurrence launches an ordinary cezar task with the workflow and agent you choose.

Create one in Automations, or ask the agent to set it up from a prompt. Preview event filters before enabling them. The list shows triggers, upcoming runs and recent outcomes; pause an automation whenever you need to.

Task dispatch

For independent pieces of work, an agent can dispatch child tasks — for example, changes in separate modules or a fresh review of a finished branch. Eac