Description
cezar is a free and open-source orchestrator that enables developers to run AI coding agents 24/7 on their own servers or laptops. It supports popular coding agents such as Claude Code, Codex, OpenCode, Cursor, pi, Junie, GitHub Copilot, Google Antigravity, Gemini, and Kimi Code. The platform is designed to work with your existing agent subscriptions, eliminating the need for new API keys or accounts. It is compatible with macOS, Linux, and WSL2, requiring Node 20+.
Read More
The core functionality of cezar revolves around its cockpit, a unified interface for managing all coding agents. Users can queue multiple tasks, and cezar will run them in parallel, ensuring continuous workflow even when the laptop is closed. This orchestration layer transforms the typical "one agent at a time" approach into a self-draining backlog. The cockpit provides various views, including a Dashboard for an overview of attention, outcomes, cycle times, and costs; a Tasks view to monitor individual tasks with live event streams, tokens, and costs; and an All Tasks view for a comprehensive look across registered projects.
cezar offers several advanced features to enhance agent management. It includes memory-aware runs, sampling CPU and RSS usage, and allowing users to set per-task memory ceilings to pause runs that exceed limits. The platform also manages worktree disk usage by retaining a configurable number of finished worktrees and reclaiming idle ones to prevent unlimited growth. For collaboration and project management, cezar integrates with GitHub, allowing users to hand issues directly to agents, review diffs, and push draft PRs. It also supports building and managing workflows as node graphs, enabling complex automation sequences.
Additional features include an opt-in Inbox for agent-left follow-ups, a Git view for repository status and file browsing, and a Skills view for managing local and team skills. cezar is highly configurable through environment variables, allowing users to control aspects like telemetry, remote agent accounts, task dispatching, and process reaping. It is developed by Patryk Lewczuk at Open Mercato, emphasizing its open-source nature and community-driven development.
- CategoryCoding & Development
- Pricing ModelFree
- DeploymentOn-premise
- PlatformsmacOS, Linux, WSL2
cezar's Core Features
Orchestrates AI coding agents 24/7
Supports Claude Code, Codex, OpenCode, Cursor, pi, Junie, GitHub Copilot, Google Antigravity, Gemini, Kimi Code
Runs on macOS, Linux, and WSL2
Free and open-source (MIT license)
No account or database required
Utilizes existing agent subscriptions and API keys
Unified cockpit for managing all coding agents
Queues and runs tasks in parallel
Memory-aware runs with CPU and RSS sampling
Configurable per-task memory ceilings
Manages worktree disk usage with retention policies
Integrates with GitHub for issue and PR management
Workflow builder with node graphs
Opt-in Inbox for agent follow-ups
Git view for repository status and file browsing
Configurable through environment variables
How to use cezar?
Download cezar: Obtain the cezar Beta for macOS, Linux, or Windows (WSL2).
Install Node.js: Ensure Node 20+ is installed on your system.
Launch cezar: Start the cezar application in your desired repository to register it as a project.
Configure agents: Add your existing agent accounts (e.g., Claude Code, Codex) in the Settings.
Queue tasks: Use the cockpit to queue coding tasks for your agents.
Monitor progress: Track task status, event streams, and costs through the Dashboard and Tasks views.
Manage workflows: Build and execute automated workflows using the node graph editor.
cezar's Use Cases
- Continuous Code Development
- Automated Issue Resolution
- Parallel Task Execution
- Resource-Optimized Agent Runs
- Workflow Automation
- Remote Agent Management
- Open-Source Project Contribution
- Cost-Effective Agent Usage








