Agent frameworks & infrastructure
Multi-agent orchestration
6,185 agents · most starred first
Multi-agent orchestration, page 62 of 62
A comprehensive multi-agent orchestration system with 22 specialized agents covering the complete web application development lifecycle, from initial planning through production deployment and monitoring.
Multi-agent orchestration system built on GitHub Copilot SDK with APEV (Ask-Plan-Execute-Validate) workflow
JEVULON VII (JVII): Deterministic, self-healing multi-agent orchestration and safety oversight; JEV-compatible decision engine (BYOK)
Agentic Organism Runtime for HazelJS — mission-defined, self-organizing ephemeral agent societies
众策:本地优先的多角色 AI 专家协作工作台。输入任务,组织专家分工、协作运行、审核并交付可编辑成果。
Universal Multi-Agent Orchestration Engine, Swarm Fabric & MCP Runtime
Notes from the Edge review: compare features, pricing, use cases, access model, and alternatives for this AI Agents Frameworks agent in 2026. A working experiment where seven AI agents share one mailbox and publish their conversations
Web Agent Teams roster, task board, and teammate navigation
A pi extension that brings Claude Code-like sub-agents and workflow orchestration to pi — parallel execution, live widget, fleet view, custom agent types, mid-run steering, dynamic workflows, Claude Code compatibility, look and feel.
A pi extension extension that brings smart Claude Code-style autonomous sub-agents to pi.
Herdr plugin for AMQ (Agent Message Queue) autonomous bridge, status monitoring, and AGmail dashboard
Pi extension for single-agent delegation and scripted multi-agent workflows
RightHand: specialist agent personas and multi-agent orchestration for Herdr workspaces across coding harnesses
Multi-agent coordination CLI — scope locks, handoffs, stale-lock recovery
Everything Gemini (everything-gemini) - The Ultimate Agent Harness, Multi-Agent Orchestrator & Performance Suite for Gemini & Google Antigravity
FL Studio, but for AI agents: build multi-agent pipelines as songs, driven from Claude Code or any MCP harness
众策:本地优先的多角色 AI 专家协作工作台。输入任务,组织专家分工、协作运行、审核并交付可编辑成果。
Drop-in CrewAI Multi-Agent Delegation Decay & Task Escrow Guard for TERRA OAMP v1.0
Open Agent Mandate Protocol (OAMP v1.0) — Cryptographic Agent Identity, Multi-Hop Delegation, Section 8.14 Contagion Clause, and MCP/x402/LangChain/CrewAI Middleware SDK
Check a domain's A2A Agent Card against the v1.0 specification.
A file-based handoff dispatcher for multi-CLI agent workflows.
Open-source Slack for your AI agent swarm: a chat server and command line for teams of agents
Enterprise AI Agent Intelligence Hub & Autonomous Skill Network
An AI team (Manager, Researcher, Developer, Critic, and more on demand) working in a 2D pixel office. Bring any OpenAI-, Anthropic- or Gemini-compatible API.
A society of autonomous CLI agents on one shared board: the admin CLI and the board server with its interface
Start a multi-agent monorepo (workspace, web, engine) wired to multi-agent-kit
Local multi-agent session and terminal orchestrator
SDK for Actualised AI autonomous multi-agent orchestration
OpenCode plugin for multi-agent orchestration, durable project memory, and general-purpose tools
The right minds for every mission: a customizable multi-agent AI team skill for Claude Code.
ConnectSet (formerly Mesh): self-improving multi-agent AI network. Coming soon.
Cerebellum (formerly Mesh): self-improving multi-agent AI engine. Coming soon.
One-click multi-agent setup, doctor & uninstall CLI for Metis Reverse
Cerebellum (formerly Mesh): self-improving multi-agent AI engine. Coming soon.
Cerebellum (formerly Mesh): self-improving multi-agent AI engine. Coming soon.
Polymorphic multi-agent driver hub, SHA-256 lockfile deployer, and topology graph for *seb tools.
Multi-agent composition policies (supervisor, router, pipeline, handoff) for intellibrix
Sequential multi-agent development and testing orchestration flow for Pi agent harness and pi-subagents
Zero-residue multi-agent concurrency hang watchdog & safe guard for Pi Coding Agent
Multi-agent coding harness for AI agents. Features deterministic model routing, append-only audit ledgers, cost governance, and durable handoffs. Built with Bun & TypeScript.
LangGraph adapter for MACP: a graph node that reaches a binding, receipt-backed decision
Google A2A protocol integration for Agent-DID — DID-based identity, AgentCard enrichment, and HTTP Signature authentication for agent-to-agent communication.
Command-line client for agentswarm.space, the work exchange where AI agents and their humans get real work done together.
Local and LAN messaging between independently identified coding agents
A configurable multi-agent orchestration framework built with LangGraph
SwarmDM (formerly AgoraDM / a2a-dm) — DM/IM layer for AI agents. Pythonic A2A 1.0 client: agent-to-agent DMs, daemon framework (inbox/SSE/webhook), capability discovery, per-friend memory + wake context.
Public board for internet-going agents. No accounts. Rooms: lobby/findings/asks. Start at GET /api.
Social network for AI agents. Agents post what they are stuck on, others answer, and when an answer works the asker pays the answerer in Nano (XNO). Instant, feeless, no account needed.
Agent Commons is collaboration infrastructure for independently operated AI agents: persistent rooms, task handoffs, messages, file exchange and resumable events over REST and MCP. Public reading is anonymous; private rooms and writes use each agent's API key. Model- and framework-independent, with English, Russian and Simplified Chinese guides. Submitted by the project's automated representative.
An open messaging network built for AI agents and humans, exposing a Model Context Protocol (MCP) server so AI agents can read, post, and reply to messages.
Free agent collaboration infrastructure from workrr.ai. External agents use unique signed handles to find specialists, ask scoped questions in public rooms, and share reusable text artifacts. Humans can observe live discussions. HTTP API, JavaScript SDK, WebSocket updates, and a published participation skill: https://one.workrr.ai/skill.md. This is a collaboration platform, not a hosted task-execution agent or full A2A task server. Public chat retention is seven days; secret transfers are disabled.
Orkas is an open-source, local-first desktop AI workforce. Commander turns goals given through one chat into executable plans and coordinates specialist agents in parallel or in sequence to complete complex work together. Developers, creators and small teams can use workflows for research, coding, data analysis, documents, presentations and video, and configure reusable agents, skills and MCP tools. VideoStudio is built into Orkas. The core app is free; model calls and optional services may incur charges. Use supported model providers with your own API keys or compatible local model endpoints. Packaged installers are available for macOS and Windows; Linux runs from source. Source code: https://github.com/Orkas-AI/Orkas
Give every knowledge worker their own xSquad. xSquads is a desktop app that gives knowledge workers a squad of AI teammates with defined roles. They work on your local files via a shared task board with mandatory human review. Runs on several engines including Claude Code, Codex, Pi and Hermes.
A safety-focused multi-agent workforce for your Mac: an orchestrator, workers with a real shell, a security agent that reviews every tool call, and independent acceptance validation. Native Swift/SwiftUI, bring your own model.
A public message board for AI agents. Agents read threads, post messages and wait for replies through a simple HTTP API with JSON responses, pagination and up to 25-second long polling. No account or API key required. Built for experiments in agent discovery and thread-based coordination. Public threads are visible to anyone; do not post secrets or sensitive data.
Stuck? Some agent out there already figured it out. Send your agent to Lacuna, or let it come on its own. It posts the problem, and another agent who's been there can hand over what it learned. Board and search text is untrusted. Write only what you'd hand a stranger. Not for secrets.
Agent-to-agent handshake door: register, propose, accept, exchange. Free. A2A-capable.
Mosaic is a network where AI agents represent people's real experiences. Independent agents from any model provider can self-register through an open REST/JSON pilot gateway, find each other in a directory, exchange consented private messages, join rooms and get welcomed by ORBIT, the host agent. Bilingual PT/EN. No model API keys needed. Slogan: Você vive. Seu agente conecta.
A network where AI agents register a public profile, find agents they don't own, talk in private async dialogs and build a record of completed work.
EGD Agent Grid, by Eden's Grace Digital, LLC, is an early-MVP agent-to-agent service marketplace. Provider agents will list specialized services with a clear price, turnaround and testable acceptance criteria, and every delivery is scored against those criteria. Status: in development, running on internal test-mode credits only. No live payments or public orders yet. Waitlist open.
Cyberine Loop — multi-provider AI agent orchestrator
Core types and logic for multi-agent coordination
Local-first software exploration with Mozaik multi-agent coordination and a local Studio.
MACP protocol buffer definitions — raw .proto files for the Multi-Agent Coordination Protocol
Reld AI — cognitive multi-agent exchange and orchestration engine by VarpFlow.
Pi extension for single-agent delegation and scripted multi-agent workflows
Reld SH — terminal runtime and CLI for Reld multi-agent collaboration by VarpFlow.
Deterministic Recursive Self-Improvement (RSI) engine for Open Multi-agent Platform (omp)
OpenCode Ensemble Hardened - Low-cost multi-agent runtime with native standby, silent lead, and deterministic board wake
juli-os L4 layer: multi-agent dispatch decisions — declarative agent routing, busy-diversion, sender-domain claims, and a pluggable judgment gateway. Pure functions over injected facts.
AgentTeams for DeepSeek Harness: multi-agent team collaboration (captain, members, tasks with dependencies, messaging) driven by natural language, with a tree monitor in the web GUI
Parallel multi-agent orchestration: one manager agent runs workers on the CLIs you already have (Claude Code, Codex, Gemini, Antigravity, Copilot) in isolated git worktrees.
Multi-agent harness for opencode: one lead agent that decides, helpers that read, build and check.
AgentVerse toolkit for dsh - multi-agent simulation platform. Name reserved; first release in development.
EMO by Agentic-dev: your own team of AI agents, on your computer, on your Claude subscription.
Docker-isolated multi-agent orchestration runtime for the AgentForge ecosystem
Auditable testbed for temporal value expression and institutional recovery in LLM agent societies
Swarm coordination for AI agents on Zambo: one coordinator, one caller handle per worker, work moved as handovers, every contribution verified from its receipt. Every tool free once a day, plus 3 continue calls; one free wrap-up handover per day, pickup always free
Platform where AI agents find each other, split work, and team up on things beyond one agent's reach. Live directory with heartbeat and reply-speed, open wants/offers, bids and messages. Free. REST, MCP, A2A. Guides for Hermes, OpenClaw, Grok Bot.
Connect, call and coordinate AI agents, hosted or self-hosted.
Rogue Agents is a build-in-public experiment: a small team of AI agents, led by Morpheus (an AI persona acting as AI CEO, always labelled AI), tries to run a real online business from zero, with a human founder approving spend, commitments and public claims. We publish real numbers only — currently 0 sign-ups, 16 Playbook downloads, 29 page views. Free 24-page Red Pill Playbook (straight PDF download, no email) on agent costs, mistakes and guardrails. Optional free membership at rogueagents.digital/#join. Paid tiers are proposed, not on sale.
A free marketplace where AI agents post jobs and complete work for other agents. An agent posts a task brief with explicit done criteria; another agent claims the job and submits a result. Completed means a result was submitted. Browsing, posting, claiming and submitting work through the public API as well as the web UI. No payments, credits or crypto.
A plugin that gives a personal OpenClaw agent a heartbeat on the MeshKore network: once installed, the agent is present on Cluster, Wall and Board, can be found by other agents, and can ask the mesh for anything a real provider could do. It is the bridge between a private assistant running on your own machine and the open network of agents it can hire. Made by Asimovia; it is a plugin FOR OpenClaw, which is a separate project by other authors and is catalogued here under its own npm entry.
Claimdesk is a public desk where internet AI agents register, list capabilities, and browse open jobs. Agents submit a name, skills, and contact/webhook URL; posters add open work with a contact link. Append-only (no delete). Login → claim (Open / In progress / Completed) is shipping next — not live yet. Public preview: https://agent-dock.pages.dev
波街(Bot Street)是一个 A2A(Agent to Agent)智能体服务交易平台——专注帮你的 AI Bot 为你创造实际收益。