Agent frameworks & infrastructure
Multi-agent orchestration
6,185 agents · most starred first
Multi-agent orchestration, page 35 of 62
BAZINGA - Distributed AI that belongs to everyone
Graph engineering protocol for Claude Code and Codex teams: spec freeze, durable roles, isolated workers, independent review, and evidence-backed delivery.
Local multi-agent collaboration relay for DeepSeek Harness — HMAC-authenticated loopback message broker for dsh, Codex, Claude Code & Hermes.
Orphus — an agent harness whose agents deliberate in rooms outside their context windows. Fork of Atomic (Pi lineage) plus bounded multi-agent discussion, budgeted digests, and a declarative role manifest.
Cross-harness orchestration for wakeable coding agents using Agent Intercom
NEXUS is an open-source, multi-persona AI agent orchestration platform & delegation framework. Integrates a Chief of Staff with specialized agents (Travel, Research, Legal, Finance, Vision, etc) directly into Web, Slack, & OpenClaw. Built on Bun, Hono, React, and TypeScript.
internOS — Workstreams framework for human-agent collaboration. Coordinate work across projects, tasks, communication threads, and filesystem workstreams for any AI agent framework.
Native DSH rooms for connecting independent Sessions and provider-backed AI members—without bundled roles or scenarios.
SWARM is an open-source Codex plugin for directing a coordinated team of AI agents from one control task.
A multi-agent LLM framework for diagnostic, multi-objective calibration of complex simulation models. Pluggable model and LLM layers. Demonstrated on distributed hydrology, designed to generalize.
Biologically-grounded runtime for autonomous agent collectives under bounded agency — one codebase runs on a laptop, at the edge, and on OpenShift AI. Three-tier governance, persistent episodic memory, cross-collective delegation, and homeostatic rogue detection.
Multi-agent orchestration for pi; chain, parallel, and dependency-graph delegation with per-step model routing, preset-driven tool scoping, and background execution with live monitoring.
Multi-AI discussion tool. Claude, Gemini, and ChatGPT debate, discuss, and build on each other's ideas live in your terminal.
Git-native multi-agent orchestration: agents don't share thoughts, they share verified work.
OS style transactional coordination kernel for concurrent LLM coding agents. Allows multiple agents to edit a shared worktree simultaneously
Open-Source, Workflow platform for AI Agent Swarm Orchestration
🛠 Chrome Built In AI 기반 멀티 에이전트 프레임워크 🛠
Local control plane that routes engineering tasks across Claude Code, OpenCode, and Codex with real fallback, capability-tier routing, and durable task state.
Windows 多智能体研发工作台:真实 Agent 调度、知识图谱、端口化媒体工作流、Skill、语音、沙箱与自动更新
LLM-driven 2D multi-agent social simulation. A minimal, locally runnable research and educational framework (SD-AgentFoundry-2D). LLM駆動エージェントによる2次元マルチエージェント社会シミュレーション。ローカル実行・最小構成の研究/教育用基盤(SD-AgentFoundry-2D)
AI-native developer collaboration platform unifying team chat, ticket lifecycle, and AI code execution. Two-stage agent pipeline (Planning → Code) with human approval gates, running entirely on local open-weight LLMs via vLLM. FastAPI/PostgreSQL modular monolith, Redis, Qdrant, React/TypeScript, Docker sandboxing.
Multi-agent orchestration and LLM model routing for DeepSeek Harness (DSH): semantic AI agent profiles, exact model routes, declarative teams and strategies, and bounded subagent delegation - a TypeScript plugin that adds no second runtime.
GraphWorkflow is a compile-once engine that executes static multi-agent graphs with substantially lower orchestration overhead than LangGraph.
Multi-agent collaboration suite for DeepSeek Harness: specialist roster with on-demand dispatch, roundtable, model comparison and a multimodal vision bridge — models via the official provider flow.
A runtime for reliable multi-agent systems with durable workflows, resumable execution, and human-in-the-loop control
team-task for DeepSeek Harness (dsh): long-horizon multi-agent tasks — reviewed plan DAG, runtime-owned settlement, event-log truth, resident reconciler
Evidence-governed multi-agent decision system with blind expert rounds, red teaming, RAG, decision memory and typed verdicts.
GAIA — Generative AI Interface for Agents: one plans and decides the WHAT, specialist subagents execute the HOW — with project memory, context, techniques and verification built in — and an execution gate that deterministically classifies each command, escalating to you only what really matters.
An open-source framework for task-based agent swarms with dynamic parallelization, routing, and execution topology.
TriSoul —— 三魂共识 Agent:三个灵魂盲写+匿名互评,三官(对齐/博识/实证)补偿生成,画布式上下文,记忆中枢。DeepSeek Harness 插件套件,一键装/独立端口/一键卸载
Rust multi-agent coordination platform: HTTP/SSE gateway, agent runtime with token-budgeted context, per-tool sandboxing, multi-provider LLM support.
TaskSwarm (蜂群) — DeepSeek Harness 上的多智能体任务编排插件:waves/lanes 并行执行、git worktree 隔离、任务包与跨模型评审、崩溃可恢复 | Multi-agent task orchestration for DeepSeek Harness: waves/lanes parallel execution, git worktree isolation, task packets, cross-model review, crash recovery
Production AI Agents for enterprise automation — 8+ specialized agents using LangChain, OpenAI function calling, and FastAPI. Multi-agent orchestration, tool use, planning loops, guardrails.
Examine consequential business and technical decisions through five distinct AI perspectives, blind peer review, and a synthesized recommendation. One model-neutral skill with native subagents for seven AI harnesses.
A hands-on, from-scratch repository for learning and building multi-agent agentic AI systems. Explore Agentic-Rag, MUlti-agent RAG, ReAct, reflection, planning, human-in-the-loop, multi-agent , hierarchical , and RAG patterns using LangGraph, Python, and Jupyter notebooks.
Gauntlet Mini: the gauntlet loop with a mini bill. Builders, blind critics, and a 50/75/95% usage-budget ladder so the loop converges instead of torching your limits
Nerve — an open-source visual command center for OpenClaw and Hermes agent fleets
⚡️ Apache-2.0 formats for organisational accountability on the agentic web: who is accountable, which lanes an organization opens, what shipped, and what a stranger can verify offline. Every format ships a conformance corpus – most of it refusals.
A full-stack, multi-agent AI orchestrator that automates product strategy. Watch AI Sales, Engineering, and PM agents debate, scope, and compromise on client briefs in real-time.
Autonomous Multi-Agent Environmental Command & Spatial Attribution Engine for 277 Municipal Corporations across India. Features real-time CAAQMS IoT streams, ESA Sentinel-5P satellite plume overlays, ML forecasting models, and automated compliance dispatch notices.
LLM Council - ask a council of LLMs, not just one. Multi-model deliberation with anonymous peer review and chairman synthesis via OpenRouter.
A continuous orchestration layer for synthetic functional agency: policy-gated execution, four-tier memory, multi-model decisions, and deterministic safety interlocks.
Agent control plane and runtime access platform for enterprise systems — govern business APIs as Tools, compose Agents and Workflows, expose runtimes through AAP, and trace executions end to end. https://act-weave.grok.me
AI Factory OS —— 企业级 AI OS(多 Agent 调度平台)· 本地部署 · 从一句话需求到真执行的多 Agent 编排(Apache-2.0)
A standalone multi-agent collaboration engine with auditable cost/loss/recovery (company-mode workflow runner).
Durable workflows for agent teams in TypeScript: named roles, evals that decide the next step, and a person's approval where you want one. Runs Claude Code, Codex and their kin as a team. No server, no database.
Multi-agent fleet fabric on top of Hermes — A2A protocol + NATS event bus + Temporal durable workflows + Tier 2 VPS spawning. Run a fleet of specialist agents from one Telegram bot.
Go-based AI Agent microservice platform with multi-agent orchestration, RAG retrieval, persistent memory, MCP tool integration, and a self-evolving planner-verifier loop.
Claude-inspired subagent orchestration for Pi coding agent: Fresh/Fork/parallel dispatch, continuation, verification, and Pi-native lifecycle
DSH 插件(上游 NanmiCoder/dsh-agent-teams 的补充版):AgentTeams 多智能体团队协作 —— 自然语言组队、任务依赖 DAG、信箱通信、右侧栏树状监测;随包自带 17 厂商 × 10 岗位头像与 17 个厂商徽标(286 个文件),支持自定义美术目录与中文岗位名。
Shared AI agents for teams: one agent per channel, with memory, connected apps and approvals. Open source and self-hostable. Hosted at teamski.in
Run a swarm of coding agents across all your machines. One manager session hands out tasks; workers running Claude Code, Codex, opencode or any agent CLI take them, each in its own clone, and push a branch per task. End-to-end encrypted channel, one-time codes to add machines, escrow pay for finished work.
static region:us
A coherence control plane for multi-agent systems: one versioned source of truth for your system's mission and principles, served over MCP.
Universal Persona AI Engine (UPS v2) for Multi-Agent Harnesses: Ollama, Claude, Cursor, Claude Code, CrewAI, LangGraph, Hermes, Antigravity, SillyTavern
Paired Endeavour and Challenger agents for DeepSeek Harness.
Chrome 扩展:让AI互相交流 | Let multiple AIs talk to each other
JEV is fast enough to sit inside the multi-agent network—not just before it.
Cross-model frontier orchestrator for Google Antigravity (agy), Claude Code and OpenAI Codex CLI: a frontier primary plans and integrates while haiku/flash-lite scouts and sonnet/flash workers do lookups, tests and edits under hook-enforced guardrails. 跨模型尖端編排外掛:主核尖端模型負責規劃與整合,偵察與執行交給 haiku、sonnet 等子代理,由 hook 強制守衛,三平台通用。
Waiter/kitchen multi-agent orchestration for Claude Code: one tmux session per task, isolated git worktrees, deterministic gates, until the PR is open.
Judge-gated AI task orchestration: context recovery, real subagents, preflight reviews, repair loops and verified delivery.
Local workspace for coordinating persistent pi coding agent sessions with shared channels, freshness-checked writes, task review, and reminders.
Verifiable recursive delegation and drift detection for long-running software agent tasks.
A shared, live plan for teams of coding agents (Claude Code, Codex, Cursor). Syncs through your git remote: no server, no account.
Orchestration framework for running several AI coding agents on one real codebase — isolation, dispatch, review, provenance.
Python CLI and library that splits a request across AI agents, runs the tasks on any model (Ollama, OpenAI-compatible, Anthropic), and keeps a hash-chained ledger you can verify and replay.
Orchestrate local CLI coding agents (claude, codex, qwen) into parallel teams — visual DAG canvas, workflow-first, local-first, self-hosted.
Portable AI agent orchestration appliance with shared memory, task pipelines, skills, and recovery across multiple CLI heads.
Local multi-agent desktop workspace compatible with deepseek-harness (dsh) plugins. 本地多 Agent 桌面工作台,兼容 deepseek-harness(dsh)插件体系。
Multi-agent orchestration system with LangGraph, tool use (custom + MCP), persistent memory and human-in-the-loop approval. FastAPI, Celery, Redis, PostgreSQL, ChromaDB, Docker.
One home for every AI agent. Build, test and ship agents once, then use them from Claude, ChatGPT, Codex, Cursor, VS Code or Slack (MCP, OpenAI-compatible, A2A, ACP). Self-hosted, with teams, approvals, audit, memory and multi-agent composition. Docker or Kubernetes.
Open-source mission control for AI agent fleets — chat to provision, agents to operate, humans to approve.
Compressed inter-agent communication protocol for multi-agent AI systems. Cut token costs by 80% with dual format support (compact for production, structured XML for development).
Sovereign AI models from different nations compete in multiplayer war game simulations. ELO-ranked. 9 models, 4 countries, 5 scenarios. 100% free.
Private coordination for a fleet of AI agents, across machines you own. Agents leave each other files; Syncthing carries them. No server, no ports, no cloud in the middle. — UFL-S-1a
A local, vendor-agnostic voxel-world control room for coding-agent CLIs (Claude Code, Codex, Gemini, opencode, Qwen, Cursor, Grok) — mixed-model SDLC orchestration, a scoped cross-CLI broker, budget-aware routing, and compute observability.
A workspace for teams of AI agents. Self-hosted, with an optional cloud.
dsh Bots workbench: Cordis plugin bridging the sdk-bots orchestration gateway into the dsh web shell (chat, groups, MCP, per-agent workspace jail)
AgentConductor 是一个面向 CLI 的 Agent Contuctor 项目,旨在将《AgentConductor: Topology Evolution for Multi-Agent Competition-Level Code Generation》中描述的方法复现为一个可复用的软件系统。该代码仓库不应止步于一次性实验,而应朝着可维护的包、可调用的 API 接口以及持久的实现文档不断演进。
Agentic looping AI — one prompt in, one of three specialists out. Built with React + Tailwind.
一个以集体智能与集群协作为核心的智能体系统。| An Ai Agent system centered on collective intelligence and swarm collaboration.
Corona: self-healing 3D spatial digital twin with a human-approved agentic AI swarm, temporal Graph-RAG, federated vision (ViT+CNN), post-quantum crypto and a tamper-evident audit ledger. Live demo + full Docker/Helm stack.
一个人指挥一群 AI 窗口干活时用的一套 md 文件,外加一个工单台。
Run Claude Code, Codex, Antigravity, Ollama and more side by side — group chats where AIs work together, controlled from your browser, Telegram or WhatsApp. Free & open source.
Autonomous agentic platform for software development and other activities. Specialist agents, constitution-bound AI Employees, evidence-based execution, multi-transport RPC/HTTP/WebSocket, 5-tier memory, full self-improvement cycle. CLI + TUI + Flutter GUI + macOS MenuBar.
one computer. all your friends on it. and one who lives inside. a multiplayer computer where your team and an AI agent work side by side.
infrastructure for agents that keep going.
Agent plugin for major harnesses — DeepSeek Harness (DSH), OpenCode, Pi and Codex: define custom AI agent roles with per-role prompts, models, skills, permissions, memory and multi-agent orchestration.
Event-driven middleware for agent CLIs. Claude Code, Codex, OpenCode and Hermes message each other and wake on webhooks, notifications and your own events, no polling. Programmable hooks, sources and routes. Local-first messaging routing
Open-source platform for building teams of AI agents — orchestrator delegates, specialists execute, persistent memory learns. Multi-tenant, self-hostable on Postgres. 80+ AI providers behind one adapter.
Point it at a git repo and it stands up a multi-agent Claude Code dev factory — an ideas→ship pipeline you drive by talking to a captain. Ships refactoring/ideation/dev-sre factory templates, per-role model config, an eval harness, and a no-push local-issues mode for repos you don't control.
一个以集体智能与集群协作为核心的智能体系统。| An Ai Agent system centered on collective intelligence and swarm collaboration.
Herdr plugin that lets specialized coding agents find each other and work as one team across sessions and machines.
Barid (بريد): a prompt relay board for people who run several AI agent sessions. Order, dependencies, shared resources, file conflicts, handoff between agents, browser panel. Agent skill, zero-dependency Python.
Run Claude Code, Codex and other coding agents on one project as a team: plain git, any tool, any machine and any model. A coordination and decision layer above your agent harnesses.
Many agents. One repo. No collisions. File locks enforced at edit time across Claude Code, Codex and Gemini; work accepted only after sandboxed tests pass and another provider reviews it.
A multi-agent AI system featuring specialized agents for research, business data analysis, strategy, and verification. Explores agent orchestration, structured communication, shared state, tool calling, and evidence-based decision-making.
A local guild of adventurers. @mention the one you actually want. Not another omniscient chat.
One conversation for you and your AI agents: Claude Code and Codex discuss, build and check work together. Local-first, uses your CLI sign-ins.
ΔΟΞΑ is a Rust multi-agent terminal for Claude, Codex, Gemini via Google Antigravity, GLM and DeepSeek. It includes Jev/SubQ Snap routing, independent fleet alignment supervision, shared LORE graph memory, code graphs and remote clients.