Awesome Agent Orchestrators

A curated list of tools and frameworks for orchestrating AI coding agents.

Parallel Agent Runners

Tools for running multiple coding agents simultaneously on different tasks.

• 1code - UI for Claude Code with local and remote agent execution.
• 5dive - Run a company of named AI agents on a server you own, each with its own model, memory, and role. Agents share an org chart and backlog, hand work to each other, and escalate to a human over Telegram. Multi-runtime (Claude Code, Codex, Grok, Antigravity, opencode).
• agent-deck - Terminal session manager for AI coding agents.
• agent-kanban - Agent-first kanban board with leader-worker model, cryptographic agent identity, and multi-runtime support (Claude Code, Codex, Gemini CLI).
• agent-of-empires - A terminal session manager for AI coding agents on Linux and macOS.
• agent-orchestrator - Agentic orchestrator for parallel coding agents.
• agentbox - Run multiple coding agents in parallel, each teleported into its own sandboxed box (local Docker or cloud VMs via Hetzner/Daytona/Vercel/E2B) with sub-1s checkpoint starts. Works with Claude Code, Codex, and OpenCode.
• agenttier - Kubernetes-native runtime that runs each AI coding agent in its own Pod + PVC sandbox, with default-deny NetworkPolicy and a streaming SSE invoke API. Run multiple agents in parallel by creating multiple Sandbox CRs.
• ai-maestro - Dashboard for orchestrating Claude, Aider, and Cursor agents across machines.
• AGX - Local-first agent orchestrator with parallel execution, wake-work-sleep checkpointing, and human-in-the-loop gates.
• aizen - macOS workspace for managing git worktrees with integrated agent sessions.
• amux - TUI for easily running parallel coding agents.
• Aperant - Autonomous multi-session AI coding.
• ariana - The IDE of the future.
• automaker - Autonomous AI development studio.
• bernstein - Deterministic orchestrator — spawns parallel AI coding agents (Claude Code, Codex CLI, Gemini CLI), verifies with tests, auto-commits. Zero LLM tokens on coordination.
• Claude Command Center (CCC) - Local dashboard for spawning, monitoring, and resuming parallel Claude Code, Codex, Cursor, Antigravity, and Kilo Code sessions.
• claude-squad - Manage multiple AI terminal agents in background.
• claude_code_bridge - Real-time multi-AI collaboration.
• clave - Native macOS app for running multiple Claude Code sessions in parallel with split/grid layouts, session groups, SSH remote sessions, and usage analytics. Local-first and MIT.
• clideck - WhatsApp-like dashboard for managing multiple AI coding agents (Claude Code, Codex, Gemini CLI, OpenCode) in one browser window. Live status, session resume, autopilot routing between agents, and full control from a phone while away.
• cmux - Open-source platform for running multiple coding agents in parallel.
• CodexMonitor - Orchestrate multiple Codex agents across local workspaces.
• CodeNomad - The command center that puts AI coding on steroids.
• collaborator - A place to create with agents.
• constellagent - macOS app for running multiple AI agents with their own terminal, editor, and git worktree.
• crystal - Run multiple Codex and Claude Code sessions in parallel git worktrees.
• dmux - Parallel agents with tmux and worktrees.
• dorothy - Desktop app to orchestrate multiple AI CLI agents with automations, Kanban management, and MCP servers.
• Emdash - Run multiple coding agents in parallel.
• ghast - Multitask with multiple terminals.
• herdr - Agent-aware terminal multiplexer with persistent workspaces, tabs, panes, and status detection for CLI coding agents.
• humanlayer - Get AI coding agents to solve hard problems in complex codebases.
• IM.codes - Mobile/web control layer for Claude Code, Codex, Gemini CLI, and other terminal-based coding agents, built for away-from-desk continuation with terminal access, file browsing, git views, localhost preview, notifications, scheduled tasks, and multi-agent workflows.
• ivy-tendril - Open-source AI coding orchestrator that manages Claude Code, Codex, Antigravity, Copilot, and OpenCode through a plan-based lifecycle with verification gates, self-improving memory, and human-in-the-loop checkpoints. (tendril.ivy.app)
• jat - The World's First Agentic IDE.
• jean - Desktop & web app for orchestrating coding agents (Claude, Codex, OpenCode) across projects and git worktrees.
• lalph - LLM agent orchestrator driven by your chosen source of issues.
• tlbx - Self-hosted browser workspace for running Claude Code, Codex, OpenCode, and other CLI agents in parallel across persistent real PTY sessions on your own machines, with control from any browser or phone.
• mux - A desktop app for isolated, parallel agentic development.
• multica - Agent-first kanban board, multi-runtime support.
• nimbalyst - The open-source visual workspace for building with Codex, Claude Code. Parallel sessions, git worktrees, kanban, visual editing.
• octomux - Local dashboard for running parallel Claude Code and Cursor agents, each in its own git worktree, with a unified permission inbox, live monitor grid, and in-app diff review. MIT.
• openkanban - TUI kanban board for orchestrating AI coding agents.
• Orca - IDE for running multiple CLI coding agents side-by-side across isolated git worktrees.
• parallel-code - Desktop app for orchestrating multiple AI coding agents (Claude Code, Codex CLI, Gemini CLI) simultaneously in isolated git worktrees with built-in diff viewer and one-click merge.
• Proliferate - Open-source local and cloud agent IDE for running Claude Code, Codex, Gemini CLI, and other coding agents in parallel across isolated workspaces.
• sortie - Turns issue tracker tickets into autonomous coding agent sessions. Agent-agnostic, tracker-agnostic, single Go binary with SQLite persistence.
• subtask - Claude Skill to do tasks with subagents in Git worktrees.
• supacode - Native macOS coding agent orchestrator.
• superset - A terminal built for coding agents.
• symphony - Turns project work into isolated, autonomous implementation runs.
• t3code - Minimal web GUI for coding agents.
• thurbox - Multi-session TUI orchestrator that runs many coding-agent CLIs (Claude Code, Codex, opencode, Aider, Copilot, and any CLI you define) in persistent tmux sessions, with git worktree isolation, remote SSH sessions, inter-session messaging, and a native code-review view. (thurbox.thurbeen.eu)
• tmux-ide - Tmux-powered terminal IDE with ide.yml layouts, agent-team templates, and Claude Code integration.
• tutti - Multi-agent orchestration CLI with config-driven workflows, git worktree isolation, and typed artifact flow between agents.
• Tura - Local-first open-source coding agent with concurrent multi-session GUI and TUI workflows, durable sessions, and OpenAI-compatible or local model support.
• vibe-kanban - Kanban board for managing AI coding agents.
• vibe-tree - Vibe code with Claude in parallel git worktrees.
• vibecraft - An RTS-style workspace for managing AI coding agents.
• repomon - A Rust TUI that spawns and supervises many AI coding agents at once across multiple repositories and git worktrees. Agents run in durable tmux sessions; the ones waiting on you float to the top, and you can approve a prompt from your phone.

Personal Assistants

AI assistants that bridge to messaging platforms and other interfaces.

• accomplish - Open source AI coworker that lives on your desktop.
• aeon - Autonomous agent framework that runs unattended on GitHub Actions; 90+ skills with quality scoring, self-healing, persistent memory, and reactive triggers.
• assistant - Panel-based personal assistant with a plugin architecture for productivity workflows.
• babyagi3 - A minimal AI agent you configure once, then run through natural language.
• cashclaw - An autonomous agent that takes work, does work, gets paid, and gets better at it.
• ClawWork - OpenClaw as your AI coworker.
• CoPaw - Your Personal AI Assistant.
• denchclaw - Managed OpenClaw framework for CRM, sales automation, and outreach agents.
• ghostclaw - An AI agent that lives on your computer and works for you.
• hermes-agent - The agent that grows with you.
• ironclaw - OpenClaw-inspired implementation in Rust focused on privacy and security.
• lemon - Local-first assistant and coding agent system.
• leon - Open-source personal assistant with voice and text interfaces.
• lettabot - Your personal AI assistant that remembers everything.
• LionClaw - Secure-first local AI assistant with durable sessions and installable skills.
• lobsterai - Your 24/7 all-scenario AI agent that gets work done for you.
• lucinate - Go-based terminal-native TUI chat client for OpenClaw, Hermes, and any OpenAI-compatible endpoint. Streaming responses, markdown rendering, tool call cards, local skills, cron management, session browsing, thinking control, and multi-agent support. Cross-platform, MIT licensed.
• mercury - Personal AI assistant that lives where you chat.
• MetaClaw - Just talk to your agent — it learns and evolves.
• nanobot - Ultra-lightweight personal AI assistant.
• nanoclaw - Lightweight alternative to OpenClaw that runs in Apple containers for security.
• NemoClaw - NVIDIA plugin for secure installation of OpenClaw.
• nullclaw - Fastest, smallest, and fully autonomous AI assistant infrastructure.
• openclaw - Your own personal AI assistant.
• piclaw - A pi-based general-purpose agent.
• picoclaw - Ultra-efficient AI assistant.
• rho - An AI agent that stays running, remembers across sessions, and checks in on its own.
• rowboat - Open-source AI coworker, with memory.
• takopi - Telegram bridge for codex, claude code, opencode, pi.
• zclaw - The smallest possible AI personal assistant for ESP32.
• zeroclaw - Fast, small, and fully autonomous AI assistant infrastructure.

Multi-Agent Swarms

Systems for coordinating multiple specialized agents working together.

• Agent Teams - Desktop app where you give high-level commands to autonomous AI agent teams across Claude, Codex, and OpenCode, and agents handle the work themselves with inter-agent messaging, Kanban task management, and built-in code review. Supports 200+ models and 75+ LLM providers.
• agentsmesh - The AI Agent Workforce Platform. Spin up remote AI workstations (AgentPods) with PTY sandboxes and git worktree isolation, coordinate multi-agent collaboration across channels and pod bindings, and manage tasks with a built-in Kanban. Self-hostable with BYOK. Supports Claude Code, Codex CLI, Gemini CLI, Aider, and OpenCode.
• antfarm - Build your agent team in OpenClaw with one command.
• automata - Agent swarming organization system.
• centaur - Self-hosted team agent platform with Slack-native conversations, Kubernetes sandboxes, shared tools, and durable workflows.
• claude-flow - Deploy multi-agent swarms with coordinated workflows.
• clawe - Multi-agent coordination system: think Trello for OpenClaw agents.
• ClawTeam - Agent Swarm Intelligence (One Command → Full Automation).
• CompanyHelm - Distributed multi-agent orchestrator with task management and agent-to-agent conversations
• Fusion - Multi-node, multi-platform agent orchestrator with a kanban board, plan-review-execute workflow gates, per-task git worktrees, and hierarchical missions.
• gastown - Multi-agent orchestration system with persistent work tracking.
• gnap - Git-Native Agent Protocol: coordinate multiple agents via a shared git repo acting as a persistent task board (todo/doing/done), no orchestrator process required.
• guild - Shared context, memory, and task coordination across AI coding agents. Single Go binary, local SQLite, hybrid keyword and semantic search.
• hcom - Hook your AI coding agents together so they can message, watch, and spawn each other across terminals.
• Hephaestus - Open Agent OS for Claude Code, Codex, and Cursor with a meta-agent builder, A2A Hub routing, local ontology, and governed memory/security gates.
• kodo - Autonomous multi-agent coding orchestrator that directs Claude Code, Codex, and Gemini CLI agents through work cycles with independent verification.
• loki-mode - Autonomous SDLC orchestrator: PRD-to-deployed-product. 41 specialized agents in 8 swarms (engineering, ops, business, data, product, growth, review, orchestration), RARV cycles (Reason-Act-Reflect-Verify), 9 quality gates, blind 3-reviewer code review, anti-sycophancy completion council. Multi-provider (Claude Code full; Codex/Gemini/Cline/Aider degraded). Local-first, open-source, BUSL-1.1.
• loom - Infrastructure for evolutionary software where autonomous loops evolve products.
• MiroShark - Swarm-intelligence engine where hundreds of grounded LLM personas coordinate across simulated Twitter, Reddit, and a prediction market to model how a population reacts to a scenario hour-by-hour. Neo4j-grounded personas, per-agent belief state, counterfactual branching, and per-agent MCP tools.
• multi-agent-shogun - Samurai-inspired tmux orchestrator with a shogun → karo → ashigaru hierarchy for running up to 10 parallel AI coding agents (Claude Code, Codex, Copilot, Kimi) with zero coordination API cost.
• openfang - Open-source Agent Operating System.
• opengoat - Build AI autonomous organizations of OpenClaw agents.
• orc - Hierarchical multi-agent orchestrator that coordinates AI coding agents through planning, task decomposition, isolated worktrees, and review pipelines.
• ORCH - CLI runtime for managing Claude Code, Codex, and Cursor as typed agent teams with state machine, goals, and TUI.
• paperclip - Orchestration for zero-human companies.
• scion - Multi-agent orchestration testbed that runs AI agents in parallel isolated containers with separate workspaces, dynamic coordination, and normalized telemetry.
• shire - Persistent workspaces for AI agent teams with inter-agent mailboxes, shared drive, and full context preservation. Supports Claude Code, OpenCode, Pi Agent and more.
• skillfold - Configuration language and compiler for multi-agent AI pipelines. Compiles YAML config into agent skills for Claude Code, Cursor, Codex, Copilot, Gemini CLI, and Windsurf.
• agent-runbook - Python CLI that compiles contract-based YAML runbooks into SKILL.md files for Claude Code and Codex. Define multi-step agent workflows with loops, branching, parallelism, and file-based state passing — write the runbook once, generate executable skills with one command.
• swarm-protocol - Headless coordination layer exposed as MCP server: claim work, detect file conflicts, heartbeat, and hand off tasks across agent sessions.
• wit - Coordination protocol that locks specific functions (not files) via Tree-sitter AST parsing. Agents declare intents, acquire symbol level locks, and get conflict warnings before writing code.
• handoff - Delegate tasks to DeepSeek V4, Codex, or Claude Opus right inside your Claude Code / Codex session. Runs in background; result returns automatically.

Autonomous Loop Runners

Projects implementing the "keep running until done" pattern.

• Agon - Autonomous research system that orchestrates scientist, coder, and auditor loops from topic to idea, proposal, and experiment.
• LoopTroop - Local GUI orchestrator for long-running AI coding tasks with LLM council planning, OpenCode execution in isolated git worktrees, and Ralph-style recovery loops that retry failed beads with fresh context.
• MartinLoop - Control plane for AI coding agents with hard budget stops, verifier gates, rollback evidence, and inspectable run receipts.
• ralph-claude-code - Autonomous AI development loop for Claude Code with intelligent exit detection.
• ralph-orchestrator - Hat-based orchestration that keeps agents in a loop until done.
• ralph-tui - Orchestrate AI coding agents to work through task lists autonomously.
• ralphy - Runs AI agents on tasks until done.
• Dex - Structured Ralph orchestrator with human-gated planning, programmatic task tracking, parallel multi-reviewer code review, automatic retries with backoff, and autonomous dead-end-aware research loops inspired by Karpathy's autoresearch; supports 7 CLI backends and ships cross-platform binaries.
• toryo - Intelligent agent orchestrator with trust-based delegation, quality ratcheting (git commit/revert), and Ralph Loop retries. Chains Claude Code, Aider, Gemini CLI, Ollama.
• wreckit - Run Ralph Wiggum Loop over your roadmap.
• neuralyzer - Allow agent to wipe its own session context and re-run the first message. Easier and more ergonomic Ralph loop engineering.