mixture-of-agents 开源项目推荐（90 个）

phj1081 /EJClaw

EYEJOKER agent stack — thin carry layer over upstream NanoClaw (host-executor MCP, gpt-proxy provider, Discord adapter)

gyunggyung /Tiny-MoA

CPU上运行的代理混合配置：LFM2.5 Brain（1.2B）+ Falcon-R Reasoner（600M）+ Tool Caller（90M）。仅支持CPU，16GB内存。轻量级AI Legion。

Burns1028 /moa-plus

Multi-agent collaboration for DeepSeek Harness: persistent agents, shared knowledge, direct communication, and independent review.

obirler /LLMProxy

LLMProxy is an intelligent large language model backend routing proxy service.

John-Wendell /Attention-MoA

Official code of Attention-MoA: Enhancing Mixture-of-Agents via Inter-Agent Semantic Attention and Deep Residual Synthesis

lac5q /beastmode

Beastmode: MofA (Mixture of Agents) orchestration framework for Hermes/OpenClaw/Codex with MemroOS-style context continuity.

winsonluk /ithy

A self-contained study of mixture-of-experts routing and agent aggregation, with offline Python experiments and synthetic fixtures.

Fu-Jie /open-webui-pipeline-wisdom-synthesizer

An external pipeline filter for Open WebUI that refactors aggregate requests with collective wisdom to output structured expert analysis reports.

GooDAnDReaDY /dsh-moa

Mixture of Agents (MoA) plugin for DeepSeek Harness with /moa slash command, file workspaces, and Live Canvas integration

jgravelle /autogroq-moa

This Streamlit application showcases the Mixture of Agents (MOA) architecture proposed by Together AI, powered by Groq LLMs. It allows users to interact with a configurable multi-agent system for enhanced AI-driven conversations.

leodenglovescode /llmgps

GPS for your LLMs

Maryam-Nasseri /MoA-local

A simplified agentic workflow process based on the Mixture of Agents (MoA) system for Large Language Models (LLMs)

tylerreckart /arbiter

A local-first, multi-agent runtime for the terminal, written in C++.

cpcc /moa

Cloudflare Workers AI 上的 Mixture-of-Agents 服务——多模型组合逼近前沿模型能力

bmtrnavsky /storm-fusion-research-pipeline

Human-gated perspective-panel research: one question, several seats, one judge. Solo, native panel, or per-seat MoA/Fusion fanout.

Xuan0629 /unison

Unison (万物一心) — A local-first, file-driven Loop Engineering pipeline where AI agents plan, build, challenge, and verify work until PASS. | 本地优先、文件驱动的多 Agent 循环工程管道。

seanyang1983 /decuria

三个臭皮匠 · 多模型智囊团 Dashboard 插件 · Mixture of Agents for Hermes Agent

raultov /opencode-moa-fusion

Mixture-of-Agents (MoA) plugin for OpenCode. It fans out a single prompt to multiple models (subagents) in parallel. The primary agent then evaluates and synthesizes the outputs into a unified, high-quality response. Child worker sessions persist and are navigable via the TUI.

Alfred1893 /dual-gate-moa

Dual-Gate-MoA: harness-enforced dual gates for LLM agents (recon before acting, QA before shipping) + MoA multi-candidate competition. Ships as a self-contained replication PRD.

Vincent-A-Yang /OninOneProxy

Self-hosted AI gateway & LLM proxy: zero-downtime failover, MOA Fusion, token-saving compression. One OpenAI-compatible endpoint for Claude Code, Cursor, Codex, Cline + 40+ providers.