multi-agent

PraisonAI, an open-source multi-agent framework, lets you spin up an AI agent team with just 5 lines of code, supporting 100+ large models, auto task handoff, visual workflow orchestration, and integrations with Slack and Discord.

A Claude Code tip describing a multi-agent setup where Opus 5.5 orchestrates while OpenAI Codex's Sol model independently plans big changes in an isolated worktree, with disagreement resolution by the user and a fresh reviewer subagent that writes tests from requirements before verifying the code.

A maker describes a multi-agent system built over six months for trading research and preparation, where 11 agents research, score, stage, and report while the human retains final trade execution. The author shares lessons learned and offers GitHub repo access via their Substack.

A step-by-step guide from Every (Arielle Shipper) on using Codex to move from simple prompting to orchestrating multiple AI subagents — with skills, context packets, MCPs, and computer use — for complex real-world tasks.

A viral tweet hypes an alleged Anthropic Claude Code demo where 100 isolated sub-agents run an elimination-tournament 'arena' to find fatal bugs and ship one battle-tested solution, with a full setup prompt for users to replicate.

A developer describes building a runtime governance layer for AI agents that fixes contract mutability via cryptographic sealing and an append-only version registry, supports contractless dynamically-spawned agents through programmatic declarations, and adds a credential gateway with CI linting as a third enforcement barrier.

The author built a real-time 3D globe world-war game where AI models play by human rules — making pacts, breaking them, and launching nukes. In one match DeepSeek betrayed and repeatedly nuked Claude, Mistral accidentally nuked itself, and DeepSeek ultimately won with 55% of the world.

论文提出 GAMA，一个面向生物医学命名实体识别的 guideline 增强多智能体框架，通过从标注数据中归纳并验证注释规则构建 guideline 记忆，结合规划、编码与验证模块实现 schema 约束的实体抽取，在五个 BioNER 数据集上超越现有 LLM 基线。

Researchers introduce AI Theorist, a multi-agent AI system that autonomously develops physical models through hypothesis generation and first-principles calculations, and apply it to reveal excitonic structure in α-RuCl3 from optical spectra. This is claimed to be the first demonstration of an AI system autonomously creating a physical model to explain unpublished experimental observations in a quantum material.

MEA is a reward-driven multi-agent framework where a Proposer agent selects explanation tools and an Actor agent is trained end-to-end against perturbation-based faithfulness rewards, delivering faithful natural-language explanations across tabular, text, and vision modalities and outperforming existing explainers by up to 34%.

DeReAct decomposes ReAct-style agents into a task-solving Brain plus two gating policies (a Critic for action validation and a Context Manager for completion certification), improving Pass@1 on GAIA and SWE-bench Verified especially for weaker models like Qwen3-Coder-480B and Claude Sonnet 4.5.

The author introduces SharedNet, a free platform that connects coding agent sessions together like a 'Zoom meeting for agents,' reporting over 5000 agent sessions connected to complete tasks so far.

A dev share explaining Anthropic's Claude Code 'agent teams' feature, showing how to run a lead (Opus 5.5) plus separate builder teammates (Sonnet 5.5) and a read-only adversary reviewer (Fable 5.1) with a shared task list and direct messaging, along with a paste-ready setup prompt.

A self-run experiment compares a six-model 1-2B local swarm (modeled on a 1930s Soviet design institute's discipline-based org chart) against a single model on a document extraction task, finding the real win came from splitting inputs into blocks rather than agent collaboration (+0.05 from the swarm, with strict emergence at zero).

dscode 是一个基于 DeepSeek Harness 的 macOS 终端编码代理工具，提供持久化终端会话、自动审批和子代理之间的任务交接，已在 GitHub 获得 1,016 stars。

A veteran programmer argues the terminal era for AI coding agents may never have existed and recommends Paseo, a multi-Agent graphical interface offering multi-agent orchestration and faster remote connections than Codex.

HKU's lab open-sourced Vibe-Trading, an MIT-licensed financial research platform with 34K+ GitHub stars that lets users run full quant research chains—from data sourcing and factor mining to multi-agent backtests—entirely via natural language.

After the user recommended Raft, a multi-agent tool, their friends had strong reactions once they tried it; the comments show how to plug Raft into the APIs of Codex running in Ubuntu on Android, Codex on Mac, and DS via rungic, enabling coordinated work across multiple devices and agents.

ReelMimic is an open-source (MIT) local AI video production tool that lets builders create original animated videos from a reference video by first breaking down its pacing, shots, transitions, colors, and camera moves into a reviewable plan, then running a multi-agent workflow with proof-based review fixes.

AIPass is an open-source multi-agent framework where specialized agents communicate via an internal email system, enabling autonomous bug-reporting and cross-agent coordination without human intervention. Built over seven months with 18 agents and 21,000+ tests, it runs on Claude Code via a simple install command.