cli-agent-orchestrator

View on GitHub

Multi-agent orchestration for AI coding CLIs — Claude Code, Kiro, Codex, and more, coordinated in isolated tmux sessions

★ 1.4K277 forksPythonApache-2.0awslabs

CAO is a Python server that orchestrates multiple AI coding CLIs (Claude Code, Codex, Kiro, Copilot, Cursor, and more) as supervisor/worker agents inside isolated tmux sessions. It adds agent profiles, MCP/HTTP control planes, scheduled flows, memory, and self-learning. POSIX-only; requires tmux 3.3+ and an authenticat

Use Cases

Run a supervisor agent that delegates work to specialist coding agents in parallelCoordinate multiple AI coding CLIs (Claude Code, Codex, Kiro, Copilot, Cursor) in isolated tmux sessionsSchedule recurring agent workflows and multi-step pipelinesGive agents persistent cross-session memory and self-learning loopsExpose agent fleet control via Web UI, CLI, MCP server, or pluginsDeploy supervisor and worker agent fleets on Kubernetes/EKSEnforce per-role tool allowlists and provider restrictionsAuthor reusable agent skills and profiles for coding tasks

Built With

Language: Python
Frameworks: FastAPI · FastMCP · MCP SDK · Pydantic · SQLAlchemy · uvicorn · libtmux · tmux · Vite · React · Playwright · APScheduler

cli-agent-orchestrator

Use Cases

Built With

Tags