Orchestrate AI Coding Agents as a Team
One command to launch a team of Claude Code, Gemini CLI, and Codex agents — coordinated through chat.
Also used to automate real business workflows beyond coding. Learn more →
See it in action
Open Source · MIT LicenseWhat is Crewly?
Crewly is an open-source platform that orchestrates Claude Code, Gemini CLI, and Codex as a development team.
Crewly is an open-source AI agent orchestration platform that coordinates multiple AI coding agents — including Claude Code, Gemini CLI, and OpenAI Codex — to work together as a software development team. It provides a real-time web dashboard for creating teams, assigning roles, delegating tasks, and monitoring agent activity in live terminal sessions. Agents communicate through bash-based skills and receive role-specific system prompts. Install with npm install -g crewly and run npx crewly start to launch the dashboard.
CrewAI is a Python framework for building LLM-powered agents through code. Crewly is a standalone AI agent orchestration platform with a web dashboard that manages existing AI coding CLIs (Claude Code, Gemini CLI, OpenAI Codex) in real terminal sessions — no Python or SDK integration required. CrewAI targets custom agent pipelines; Crewly targets developers who want to run a team of AI coding agents immediately with a single command (npx crewly start), complete with task management, persistent memory, and a knowledge base.
Crewly supports three AI coding agent runtimes: Claude Code (Anthropic, default), Gemini CLI (Google), and Codex CLI (OpenAI). Each agent runs in its own PTY terminal session. You can mix runtimes within a single team — for example, use Claude Code for backend development and Gemini CLI for frontend work.
Supports Multiple AI Runtimes
Mix runtimes within a single team
Built for Multi-Agent Development
Multi-Runtime Agents
Orchestrate teams with Claude Code, Gemini CLI, or Codex. Mix runtimes per agent — use Claude for backend and Gemini for frontend in the same team.
Real-Time Dashboard
Watch every agent work in live terminal sessions. Monitor progress, view task status, and intervene when needed — all from one browser tab.
Task Orchestration
Create projects, delegate tasks, and track progress. The orchestrator agent breaks down work and assigns it to the right team members automatically.
Knowledge Base
Store SOPs, architecture docs, and runbooks that agents can query on demand. Global and per-project scopes keep context organized and accessible.
Persistent Memory
Agents remember patterns, decisions, and gotchas across sessions. Learnings compound over time, making your AI team smarter with every task.
Skills Marketplace
Extend agent capabilities with community-built skills, roles, and configurations. Browse and install from the marketplace with a single command.
Three Steps to Your First AI Team
Install & Launch
Run npx crewly start to launch the backend and open the web dashboard.
$ npx crewly startAssemble Your Team
Pick agent roles (developer, QA, PM, architect) and choose runtimes (Claude Code, Gemini CLI, or Codex).
Delegate & Ship
Assign your team to a project, delegate tasks, and watch your AI agents collaborate in real time.
Frequently Asked Questions
Crewly is an open-source AI agent orchestration platform. It coordinates multiple AI coding agents — Claude Code, Gemini CLI, and OpenAI Codex — to work as a software development team. You manage everything through a real-time web dashboard: create teams, assign roles, delegate tasks, and watch agents work in live terminal sessions.
Crewly supports three agent runtimes: Claude Code (Anthropic), Gemini CLI (Google), and Codex CLI (OpenAI). Claude Code is the default. You can mix different runtimes within a single team and customize the launch command for each one in the settings.
Install Crewly with "npm install -g crewly", then run "npx crewly start". This launches the backend server and opens the web dashboard in your browser. You need Node.js v20+ and at least one AI coding CLI (Claude Code, Gemini CLI, or Codex) installed on your machine.
Yes. Crewly is fully open source under the MIT license. The source code is available on GitHub.
CrewAI is a Python framework for building custom LLM agent pipelines through code. Crewly is a standalone platform that orchestrates existing AI coding CLIs (Claude Code, Gemini CLI, Codex) through a web dashboard — no Python or SDK required. Crewly is designed for developers who want to launch a team of AI agents immediately, with built-in task management, persistent memory, and a knowledge base.
Yes. Install Gemini CLI ("npm install -g @google/gemini-cli"), set your GEMINI_API_KEY environment variable, and select "Gemini CLI" as the runtime when creating a team in the Crewly dashboard. Crewly launches Gemini agents with "gemini --yolo" by default.
Yes. Install the Codex CLI ("npm install -g @openai/codex"), configure your OpenAI API key, and select "OpenAI Codex" as the runtime in the dashboard. Crewly launches Codex agents with "codex --full-auto" by default.
Crewly runs on macOS, Linux, and Windows. It uses node-pty for terminal session management, which provides cross-platform PTY support. You need Node.js v20 or later installed.
Crewly includes built-in role prompts for: developer, frontend developer, backend developer, fullstack developer, architect, QA engineer, product manager, designer, generalist, and more. Each role has a tailored system prompt that guides the agent's behavior. You can also create custom roles.
Agents use bash-based skills to interact with the Crewly backend. They can send messages to specific teammates, report progress, recall memories, query the knowledge base, and manage tasks. An orchestrator agent can delegate tasks, start or stop other agents, and coordinate the entire team.
Start Orchestrating in Seconds
One command. Multiple AI agents. Real-time collaboration.