[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-74034":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":23,"hasPages":25,"topics":26,"createdAt":10,"pushedAt":10,"updatedAt":27,"readmeContent":28,"aiSummary":29,"trendingCount":16,"starSnapshotCount":16,"syncStatus":30,"lastSyncTime":31,"discoverSource":32},74034,"ClawTeam","HKUDS\u002FClawTeam","HKUDS","\"ClawTeam: Agent Swarm Intelligence\" (One Command → Full Automation)","",null,"Python",5317,727,40,13,0,16,47,158,48,39.59,"MIT License",false,"main",true,[],"2026-06-12 02:03:21","\u003Ch1 align=\"center\">\u003Cimg src=\"assets\u002Ficon.png\" alt=\"\" width=\"64\" style=\"vertical-align: middle;\">  ClawTeam: Agent Swarm Intelligence\u003C\u002Fh1>\n\n\u003Cp align=\"center\">\n  \u003Cstrong>The Evolution of AI Agents 🚀: Solo 🤖 → Swarm 🦞🤖🤖🤖\u003Cbr>\n  ClawTeam: Let AI Agents Form Swarms, Think & Work Together, and Ship Faster\u003C\u002Fstrong>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"#-quick-start\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FQuick_Start-3_min-blue?style=for-the-badge\" alt=\"Quick Start\">\u003C\u002Fa>\n  \u003Ca href=\"#-use-cases\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FUse_Cases-3_Demos-green?style=for-the-badge\" alt=\"Use Cases\">\u003C\u002Fa>\n  \u003Ca href=\"#-features\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFeatures-12+-purple?style=for-the-badge\" alt=\"Features\">\u003C\u002Fa>\n  \u003Ca href=\"LICENSE\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-yellow?style=for-the-badge\" alt=\"License\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-≥3.10-blue?logo=python&logoColor=white\" alt=\"Python\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ftyper-CLI-green\" alt=\"Typer\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fagents-Claude_Code_%7C_Codex_%7C_Any_CLI-blueviolet\" alt=\"Agents\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ftransport-File_%7C_ZeroMQ_P2P-orange\" alt=\"Transport\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FHKUDS\u002F.github\u002Fblob\u002Fmain\u002Fprofile\u002FREADME.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFeishu-Group-E9DBFC?style=flat&logo=feishu&logoColor=white\" alt=\"Feishu\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FHKUDS\u002F.github\u002Fblob\u002Fmain\u002Fprofile\u002FREADME.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FWeChat-Group-C5EAB4?style=flat&logo=wechat&logoColor=white\" alt=\"WeChat\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n**One Command Line: Full Automation.** — agents spawn swarms, delegate tasks, and deliver results. \n\nHuman provides the goal. The Agent Team orchestrates everything else.\n\nFull compatibility with [Claude Code](https:\u002F\u002Fclaude.ai\u002Fclaude-code), [Codex](https:\u002F\u002Fopenai.com\u002Fcodex), [OpenClaw](https:\u002F\u002Fgithub.com\u002Fopenclaw\u002Fopenclaw), [nanobot](https:\u002F\u002Fgithub.com\u002FHKUDS\u002Fnanobot), [Cursor](https:\u002F\u002Fcursor.com), and any CLI agent.  [**中文文档**](README_CN.md) | [**한국어**](README_KR.md)\n\n---\n\n## 📰 News\n\n**2026-03-18** ClawTeam project launched publicly.\n\n**2026-03-23** ClawTeam `v0.2.0` is released today.\n\n**2026-03** The current baseline includes config management, multi-user workflows, Web UI, P2P transport, and team templates.\n\n---\n\n## ✨ ClawTeam's Key Features\n\n\u003Ctable align=\"center\" width=\"100%\">\n\u003Ctr>\n\u003Ctd width=\"25%\" align=\"center\" style=\"vertical-align: top; padding: 15px;\">\n\n\u003Ch3>🔬 AI Research Automation\u003C\u002Fh3>\n\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAutoResearch-FF6B6B?style=for-the-badge&logo=pytorch&logoColor=white\" alt=\"AutoResearch\" \u002F>\n\u003C\u002Fdiv>\n\n\u003Cimg src=\"assets\u002Fscene-autoresearch.png\" width=\"180\">\n\n\u003Cp align=\"center\">\u003Cstrong>• Large-Scale Automated ML Experimentation\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• AI Model Training & Optimization\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• AI-Driven Hypothesis Generation & Validation\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Self-Improving Model Architectures\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003C\u002Ftd>\n\u003Ctd width=\"25%\" align=\"center\" style=\"vertical-align: top; padding: 15px;\">\n\n\u003Ch3>🏗️ Agentic Engineering\u003C\u002Fh3>\n\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FFull--Stack_Dev-4ECDC4?style=for-the-badge&logo=git&logoColor=white\" alt=\"Engineering\" \u002F>\n\u003C\u002Fdiv>\n\n\u003Cimg src=\"assets\u002Fscene-engineering.png\" width=\"180\">\n\n\u003Cp align=\"center\">\u003Cstrong>• Autonomous Full-Stack Development\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Self-Evolving Software\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Collaborative Open Source Development\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Real-Time System Integration\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003C\u002Ftd>\n\u003Ctd width=\"25%\" align=\"center\" style=\"vertical-align: top; padding: 15px;\">\n\n\u003Ch3>💰 AI Hedge Fund\u003C\u002Fh3>\n\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FInvestment_Analysis-FFD93D?style=for-the-badge&logo=bitcoin&logoColor=black\" alt=\"Hedge Fund\" \u002F>\n\u003C\u002Fdiv>\n\n\u003Cimg src=\"assets\u002Fscene-hedgefund.png\" width=\"180\">\n\n\u003Cp align=\"center\">\u003Cstrong>• Automated Market Research & Data Mining\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Multi-Strategy Portfolio Optimization\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Real-Time Risk Assessment\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Algorithmic Trading Execution & Monitoring\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003C\u002Ftd>\n\u003Ctd width=\"25%\" align=\"center\" style=\"vertical-align: top; padding: 15px;\">\n\n\u003Ch3>🎪 Your Own Swarm\u003C\u002Fh3>\n\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FTOML_Templates-C77DFF?style=for-the-badge&logo=toml&logoColor=white\" alt=\"Templates\" \u002F>\n\u003C\u002Fdiv>\n\n\u003Cimg src=\"assets\u002Fscene-template.png\" width=\"180\">\n\n\u003Cp align=\"center\">\u003Cstrong>• Custom Scientific Research Teams\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Personalized Investment Committees\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Business Operations Teams\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003Cp align=\"center\">\u003Cstrong>• Content Production Studios\u003C\u002Fstrong>\u003C\u002Fp>\n\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n---\n\n\u003Ctable align=\"center\" width=\"100%\">\n\u003Ctr>\n\u003Ctd width=\"50%\" align=\"center\" style=\"vertical-align: top; padding: 10px;\">\n\n\u003Cstrong>v0.1.0\u003C\u002Fstrong>\n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F7e2f0ecd-8fe3-4970-90ac-5c9669ff060c\n\n\u003C\u002Ftd>\n\u003Ctd width=\"50%\" align=\"center\" style=\"vertical-align: top; padding: 10px;\">\n\n\u003Cstrong>v0.2.0\u003C\u002Fstrong>\n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Ffd23be91-5cf4-457c-a77e-bac24b76e58f\n\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n\n☝️ Intelligent leader agent orchestrates 8 specialized sub-agents across 8 H100 GPUs, autonomously designing experiments and dynamically reallocating resources based on real-time performance.\n\n🧠 The system synthesizes breakthroughs across teams and evolves strategies independently — achieving full research automation without human intervention.\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Fteaser.png\" alt=\"ClawTeam - AI agents orchestrating themselves\" width=\"800\">\n\u003C\u002Fp>\n\n---\n\n## 🤔 Why ClawTeam?\n\nCurrent AI agents are powerful — but they work in **isolation**. When facing complex tasks, you're stuck manually coordinating multiple agents, juggling context, and stitching together fragmented results.\n\n**What if agents could think and work as a team?**\n\nClawTeam unlocks **Agent Swarm Intelligence** — where AI agents self-organize into collaborative teams, intelligently divide complex work, share insights in real-time, and converge on breakthrough solutions.\n\n• **🚀 Spawns specialized sub-agents** — each with dedicated environments and focus areas\n\n• **📋 Designs intelligent task allocation** — with smart dependency management\n\n• **💬 Facilitates real-time coordination** — seamless inter-agent communication\n\n• **📊 Monitors team performance** — tracks progress and identifies bottlenecks\n\n• **🔄 Adapts strategies dynamically** — reallocates resources and redirects efforts\n\n#### ✨ The Result?\nYou set the vision. The swarm executes with collective intelligence.\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Fcomic-how-it-works.png\" alt=\"How ClawTeam works - comic\" width=\"700\">\n\u003C\u002Fp>\n\n---\n\n## 🎯 Swarm Intelligence in Action\n\n\u003Ctable>\n\u003Ctr>\n\u003Ctd width=\"33%\">\n\n### 🦞 Agents Spawn Agents\nThe leader agent calls `clawteam spawn` to create workers. Each worker gets its own **git worktree**, **tmux window**, and **identity** — automatically.\n\n```bash\n# The leader agent runs:\nclawteam spawn --team my-team \\\n  --agent-name worker1 \\\n  --task \"Implement auth module\"\n```\n\n\u003C\u002Ftd>\n\u003Ctd width=\"33%\">\n\n### 🤖 Agents Talk to Agents\nWorkers check their inbox, update task status, and report results — all through CLI commands that are **auto-injected** into their prompt.\n\n```bash\n# A worker agent checks tasks:\nclawteam task list my-team --owner me\n# Then reports back:\nclawteam inbox send my-team leader \\\n  \"Auth done. All tests passing.\"\n```\n\n\u003C\u002Ftd>\n\u003Ctd width=\"33%\">\n\n### 👀 You Just Watch\nMonitor the swarm from a tiled tmux view or a Web UI. The leader handles coordination — you intervene only when you want to.\n\n```bash\n# Watch all agents simultaneously\nclawteam board attach my-team\n# Or open the web dashboard\nclawteam board serve --port 8080\n```\n\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n| | ClawTeam | Other multi-agent frameworks |\n|---|---------|----------------------------|\n| 🎯 **Who uses it** | **The AI agents themselves** | Humans writing orchestration code |\n| ⚡ **Setup** | `pip install` + one prompt to the leader | Docker, cloud APIs, YAML configs |\n| 🏗️ **Infrastructure** | Just a filesystem and tmux | Redis, message queues, databases |\n| 🤖 **Agent support** | Any CLI agent (Claude Code, Codex, OpenClaw, custom) | Framework-specific only |\n| 🌳 **Isolation** | Git worktrees (real branches, real diffs) | Containers or virtual envs |\n| 🧠 **Intelligence** | Swarm self-organizes via CLI commands | Hard-coded orchestration logic |\n\n---\n\n## 🎬 Use Cases\n\n### 🔬 1. Autonomous ML Research — 8 Agents × 8 H100 GPUs\n\nBased on [@karpathy's autoresearch](https:\u002F\u002Fgithub.com\u002Fkarpathy\u002Fautoresearch).\n\n#### 💫 One Command. Full Automation.\n\n#### Human input: \"Optimize this LLM training setup using 8 GPUs\"\n\nThe Agent Team handles everything else:\n- Spawns 8 specialized research agents across H100s\n- Designs 2000+ autonomous experiments\n- Achieves breakthrough improvements (val_bpb: 1.044→0.977)\n- Zero human intervention required\n\n#### 🎯 Pure Research at Scale\n\nTransform months of manual hyperparameter tuning into hours of intelligent automation.\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Fautoresearch-progress.png\" alt=\"AutoResearch Progress\" width=\"720\">\n  \u003Cbr>\n  \u003Cem>🏆 val_bpb: 1.044 → 0.977 (6.4% improvement) | 2430+ experiments | ~30 GPU-hours\u003C\u002Fem>\n\u003C\u002Fp>\n\n**What agent team did autonomously:**\n\n```\nHuman prompt: \"Use 8 GPUs to optimize train.py. Read program.md for instructions.\"\n\n🦞 Leader agent's actions:\n├── 📖 Read program.md, understand the experiment protocol\n├── 🏗️ clawteam team spawn-team autoresearch\n├── 🚀 Assigned each GPU a research direction:\n│   ├── GPU 0: clawteam spawn --task \"Explore model depth (DEPTH 10-16)\"\n│   ├── GPU 1: clawteam spawn --task \"Explore model width (ASPECT_RATIO 80-128)\"\n│   ├── GPU 2: clawteam spawn --task \"Tune learning rates and optimizer\"\n│   ├── GPU 3: clawteam spawn --task \"Explore batch size and accumulation\"\n│   ├── GPU 4-7: clawteam spawn tmux codex --task \"...\"  (Codex agents)\n│   └── 🌳 Each agent: own git worktree, own branch, isolated experiments\n├── 🔄 Every 30 minutes, checked results:\n│   ├── clawteam board show autoresearch\n│   ├── Read each agent's results.tsv\n│   ├── 🏆 Identified best findings (depth=12, batch=2^17, norm-before-RoPE)\n│   └── 📡 Cross-pollinated: told new agents to start from the best config\n├── 🔧 When agents finished, reassigned GPUs:\n│   ├── Killed idle agents, cleaned worktrees\n│   ├── Created new worktrees from the best commit\n│   └── Spawned fresh agents with combined optimization directions\n└── ✅ After 2430+ experiments: val_bpb 1.044 → 0.977\n```\n\nFull results: [novix-science\u002Fautoresearch](https:\u002F\u002Fgithub.com\u002Fnovix-science\u002Fautoresearch)\n\n---\n\n### 🏗️ 2. Agentic Software Engineering\n\nYou tell Claude Code: *\"Build me a full-stack todo app.\"* Claude realizes this is a multi-module task and **self-organizes a team**:\n\n```\nHuman prompt: \"Build a full-stack todo app with auth, database, and React frontend.\"\n\n🦞 Leader agent's actions:\n├── 🏗️ clawteam team spawn-team webapp -d \"Full-stack todo app\"\n├── 📋 Created tasks with dependency chains:\n│   ├── T1: \"Design REST API schema\"          → architect\n│   ├── T2: \"Implement JWT auth\" --blocked-by T1  → backend1\n│   ├── T3: \"Build database layer\" --blocked-by T1 → backend2\n│   ├── T4: \"Build React frontend\"             → frontend\n│   └── T5: \"Integration tests\" --blocked-by T2,T3,T4 → tester\n├── 🚀 Spawned 5 sub-agents (each in its own git worktree):\n│   ├── clawteam spawn --agent-name architect --task \"Design the API schema\"\n│   ├── clawteam spawn --agent-name backend1  --task \"Implement JWT auth\"\n│   ├── clawteam spawn --agent-name backend2  --task \"Build PostgreSQL models\"\n│   ├── clawteam spawn --agent-name frontend  --task \"Build React UI\"\n│   └── clawteam spawn --agent-name tester    --task \"Write pytest tests\"\n├── 🔗 Dependency auto-resolution:\n│   ├── architect completes → backend1 and backend2 auto-unblock\n│   ├── All backends complete → tester auto-unblocks\n│   └── Each agent calls: clawteam task update \u003Cid> --status completed\n├── 💬 Sub-agents coordinate via inbox:\n│   ├── architect → backend1: \"Here's the OpenAPI spec: ...\"\n│   ├── backend1 → tester: \"Auth endpoints ready at \u002Fapi\u002Fauth\u002F*\"\n│   └── tester → leader: \"All 47 tests passing ✅\"\n└── 🌳 Leader merges all worktrees into main branch\n```\n\n---\n\n### 💰 3. AI Hedge Fund — One-Command Team Launch\n\nA pre-built TOML template spawns a complete **7-agent** investment analysis team:\n\n```bash\n# One command launches everything:\nclawteam launch hedge-fund --team fund1 --goal \"Analyze AAPL, MSFT, NVDA for Q2 2026\"\n```\n\n```\n🦞 What happens automatically:\n├── 📊 Portfolio Manager (leader) spawns and receives the goal\n├── 🤖 5 Analyst agents spawn, each with a different strategy:\n│   ├── 🎩 Buffett Analyst    → value investing (moat, ROE, DCF)\n│   ├── 🚀 Growth Analyst     → disruption (TAM, network effects)\n│   ├── 📈 Technical Analyst  → indicators (EMA, RSI, Bollinger)\n│   ├── 📋 Fundamentals       → financial ratios (P\u002FE, D\u002FE, FCF)\n│   └── 📰 Sentiment Analyst  → news + insider trading signals\n├── 🛡️ Risk Manager spawns, waits for all analyst signals:\n│   ├── clawteam inbox receive fund1 (collects all 5 signals)\n│   ├── Consolidates + computes position limits\n│   └── clawteam inbox send fund1 portfolio-manager \"RISK REPORT: ...\"\n└── 💼 Portfolio Manager makes final buy\u002Fsell\u002Fhold decisions\n```\n\nTemplates are TOML files — **create your own team archetypes** for any domain.\n\n---\n\n## 📦 Install\n\n```bash\npip install clawteam\n\n# Or install into an isolated user venv at ~\u002F.clawteam\u002F.venv\nscripts\u002Finstall_clawteam.sh\n\n# Or from source\ngit clone https:\u002F\u002Fgithub.com\u002FHKUDS\u002FClawTeam.git\ncd ClawTeam\npip install -e .\n\n# Or install this checkout into ~\u002F.clawteam\u002F.venv\nscripts\u002Fclawteam_local_install\n\n# Optional: P2P transport (ZeroMQ)\npip install -e \".[p2p]\"\n```\n\nRequires **Python 3.10+**, **tmux**, and a CLI coding agent (e.g. `claude`, `codex`). The install scripts create\u002Freuse `~\u002F.clawteam\u002F.venv`, link `~\u002F.local\u002Fbin\u002Fclawteam`, and install\u002Fupdate the bundled `clawteam` skill for detected clients.\n\nSkill install targets are detected from existing config directories, including Claude Code, Codex, Gemini CLI, OpenClaw, OpenCode, Nanobot, Cursor, OpenHarness (`~\u002F.openharness\u002Fskills`), and Ohmo (`~\u002F.ohmo\u002Fskills`).\n\nAll `spawn` examples assume the agent CLI you name is already installed and available on `PATH`.\n\n---\n\n## 🚀 Quick Start\n\nIf you're new to ClawTeam, follow this order:\n\n1. Make sure `tmux` and your agent CLI run standalone on this machine.\n2. Pick one path below: let an agent drive, or drive it manually.\n3. Use the supported-agent table to choose the right `spawn` command.\n4. If you're integrating a new agent, check the adapter notes before debugging.\n\n### ✅ Before You Start\n\nRun these checks first:\n\n```bash\ntmux -V\nclawteam --help\n\n# Replace claude with the agent you actually want to use:\nclaude --version\ncodex --version\nnanobot --help\n```\n\nIf the agent CLI does not run correctly by itself, `clawteam spawn` will not fix it.\n\n### ⚡ Option 1: Let the Agent Drive (Recommended)\n\nClawTeam ships with a reusable skill in `skills\u002Fclawteam\u002F`.\n\n**Claude Code**\n\nInstall the skill into `~\u002F.claude\u002Fskills\u002Fclawteam`, then prompt:\n\n```\n\"Build a web app. Use clawteam to split the work across multiple agents.\"\n```\n\n**Codex**\n\nInstall the same skill into `$CODEX_HOME\u002Fskills\u002Fclawteam` (typically `~\u002F.codex\u002Fskills\u002Fclawteam`), then prompt:\n\n```\nUse $clawteam to split this task across multiple agents and coordinate the team to completion.\n```\n\nThe agent will automatically create a team, spawn workers, assign tasks, and coordinate — using `clawteam` CLI commands under the hood.\n\n### 🔧 Option 2: Drive It Manually\n\n```bash\n# 1. Create a team (you become the leader)\nclawteam team spawn-team my-team -d \"Build the auth module\" -n leader\n\n# 2. Spawn worker agents — each gets a git worktree, tmux window, and identity\nclawteam spawn --team my-team --agent-name alice --task \"Implement the OAuth2 flow\"\nclawteam spawn --team my-team --agent-name bob   --task \"Write unit tests for auth\"\n\n# 3. Workers auto-receive a coordination prompt that teaches them to:\n#    ✅ Check tasks:    clawteam task list my-team --owner alice\n#    ✅ Update status:  clawteam task update my-team \u003Cid> --status completed\n#    ✅ Message leader: clawteam inbox send my-team leader \"Done!\"\n#    ✅ Report idle:    clawteam lifecycle idle my-team\n\n# 4. Watch them work side-by-side\nclawteam board attach my-team\n```\n\n### 🧩 Profiles and Presets\n\nWhen you want to use a non-default provider, model, or API gateway, configure a\n**profile** first instead of manually exporting provider env vars each time.\n\n```bash\n# See built-in provider templates\nclawteam preset list\nclawteam preset show moonshot-cn\n\n# Generate a reusable runtime profile from a preset\nclawteam preset generate-profile moonshot-cn claude --name claude-kimi\n\n# MiniMax (M2.7) — global or China endpoint\nclawteam preset generate-profile minimax-global claude --name claude-minimax\nclawteam preset generate-profile minimax-cn claude --name claude-minimax-cn\n\n# Or use the interactive TUI\nclawteam profile wizard\n\n# Claude Code on a fresh machine\u002Fhome may need this once\nclawteam profile doctor claude\n\n# Smoke-test the profile before spawning workers\nMOONSHOT_API_KEY=... clawteam profile test claude-kimi\nMINIMAX_API_KEY=... clawteam profile test claude-minimax\n```\n\nRules of thumb:\n\n- `profile` is the final runtime object used by `spawn` \u002F `launch`\n- `preset` is a reusable provider template that generates one or more profiles\n- `wizard` is the easiest path for first-time setup\n- `doctor` is mainly for Claude Code first-run onboarding state\n\n### 🧭 Which Spawn Command Should I Use?\n\nUse `clawteam spawn [backend] [command] ...` with the command that already works on\nyour machine:\n\n```bash\n# Claude Code\nclawteam spawn tmux claude --team my-team --agent-name alice --task \"Implement OAuth2\"\n\n# Codex\nclawteam spawn tmux codex --team my-team --agent-name bob --task \"Write frontend tests\"\n\n# nanobot\nclawteam spawn tmux nanobot --team my-team --agent-name carol --task \"Build the API\"\n\n# A configured profile (recommended for non-default providers\u002Fmodels)\nclawteam spawn tmux --profile claude-kimi --team my-team --agent-name dave --task \"Refactor the auth flow\"\n```\n\nNotes:\n\n- `tmux` is the default backend and is the best choice when you want to watch interactive agent UIs.\n- `subprocess` is better for one-shot tools or non-interactive scripts.\n- `nanobot` is normalized internally to `nanobot agent`, so the command above is the correct ClawTeam entrypoint.\n- Claude Code and Codex trust prompts in fresh worktrees are auto-confirmed by the tmux backend.\n- For non-default providers\u002Fmodels, prefer `--profile \u003Cname>` over manually exporting env vars inline.\n\n### 🔌 Adding a Different Agent\n\nClawTeam can work with agents beyond Claude Code, Codex, and nanobot, but the CLI\nmust satisfy a small compatibility contract:\n\n1. The command must exist on `PATH` and launch successfully outside ClawTeam.\n2. The agent must be able to run inside a specific working directory or git worktree.\n3. The agent must accept an initial task, either by command-line argument or interactive input.\n4. The process must stay alive in `tmux` if it is meant to be interactive.\n\nIf you're unsure, test the agent standalone first, then wrap it with:\n\n```bash\nclawteam spawn subprocess \u003Cyour-agent> --team my-team --agent-name test --task \"Say OK\"\n```\n\nIf that works, switch to `tmux` for interactive monitoring.\n\n### 🤖 Supported Agents\n\nClawTeam works with **any CLI agent** that can execute shell commands:\n\nAll examples below assume the corresponding CLI already runs standalone on your machine.\n\n| Agent | Spawn Command | Status |\n|-------|--------------|--------|\n| [Claude Code](https:\u002F\u002Fclaude.ai\u002Fclaude-code) | `clawteam spawn tmux claude --team ...` | ✅ Full support |\n| [Codex](https:\u002F\u002Fopenai.com\u002Fcodex) | `clawteam spawn tmux codex --team ...` | ✅ Full support |\n| [OpenClaw](https:\u002F\u002Fgithub.com\u002Fopenclaw\u002Fopenclaw) | `clawteam spawn tmux openclaw --team ...` | ✅ Full support |\n| [nanobot](https:\u002F\u002Fgithub.com\u002FHKUDS\u002Fnanobot) | `clawteam spawn tmux nanobot --team ...` | ✅ Full support |\n| [Kimi CLI](https:\u002F\u002Fgithub.com\u002FMoonshotAI\u002Fkimi-cli) | `clawteam spawn tmux kimi --team ...` | ✅ Full support |\n| [Cursor](https:\u002F\u002Fcursor.com) | `clawteam spawn subprocess cursor --team ...` | 🔮 Experimental |\n| Custom scripts | `clawteam spawn subprocess python --team ...` | ✅ Full support |\n\nFor provider-aware setups such as Claude Code via Moonshot Kimi, MiniMax, or Gemini via\nVertex, use `profile` + `preset` and then spawn with `--profile`.\n\n---\n\n## ✨ Features\n\n\u003Ctable>\n\u003Ctr>\n\u003Ctd width=\"50%\">\n\n### 🦞 Agent Self-Organization\n- Leader agents spawn and manage worker agents\n- **Auto-injected coordination prompt** — zero manual setup\n- Workers self-report status, results, and idle state\n- Works with any CLI agent: Claude Code, Codex, OpenClaw, custom\n\n### 🌳 Workspace Isolation\n- Each agent gets its own **git worktree** (separate branch)\n- No merge conflicts between parallel agents\n- Checkpoint, merge, and cleanup commands\n- Branch naming: `clawteam\u002F{team}\u002F{agent}`\n\n### 📋 Task Tracking with Dependencies\n- Shared kanban: `pending` → `in_progress` → `completed` \u002F `blocked`\n- `--blocked-by` dependency chains — **auto-unblock on completion**\n- `task wait` blocks until all tasks complete\n- Filter by status, owner; JSON output for scripting\n\n\u003C\u002Ftd>\n\u003Ctd width=\"50%\">\n\n### 💬 Inter-Agent Messaging\n- Point-to-point **inboxes** (send, receive, peek)\n- **Broadcast** to all team members\n- File-based (default) or ZeroMQ P2P transport with offline fallback\n- Agents discover messages via `inbox receive`\n\n### 📊 Monitoring & Dashboards\n- `board show` — terminal kanban board\n- `board live` — auto-refreshing dashboard\n- `board attach` — **tiled tmux view** of all agents working\n- `board serve` — **Web UI** with real-time updates\n\n### 🎪 Team Templates\n- **TOML files** define team archetypes (roles, tasks, prompts)\n- One command launches a complete team: `clawteam launch \u003Ctemplate>`\n- Built-in: AI Hedge Fund (7 agents). Create your own.\n- Variable substitution: `{goal}`, `{team_name}`, `{agent_name}`\n\n\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### 🔧 More\n\n| Feature | Description |\n|---------|-------------|\n| 📝 **Plan Approval** | Agents submit plans for leader review before execution |\n| 🔄 **Lifecycle Protocol** | Graceful shutdown request\u002Fapprove\u002Freject, idle notifications |\n| 📊 **JSON Output** | `--json` flag on all commands — agents parse structured output |\n| 🌐 **Cross-Machine** | Shared filesystem (NFS\u002FSSHFS) or P2P transport for distributed teams |\n| 👥 **Multi-User** | Namespace agents by user — multiple humans can share a team |\n| ⚙️ **Configuration** | Persistent config: env var > config file > default priority |\n| 🔌 **Agent Skill** | Reusable skill entry for Claude Code and Codex workflows |\n\n---\n\n## 🤖 How Agents Use ClawTeam\n\nWhen an agent is spawned via `clawteam spawn`, it receives an **auto-injected coordination prompt**:\n\n```\n## Coordination Protocol (auto-injected into every spawned agent)\n\n- 📋 Check your tasks: clawteam task list \u003Cteam> --owner \u003Cyour-name>\n- ▶️ Start a task:     clawteam task update \u003Cteam> \u003Cid> --status in_progress\n- ✅ Finish a task:    clawteam task update \u003Cteam> \u003Cid> --status completed\n- 💬 Message leader:   clawteam inbox send \u003Cteam> leader \"status update...\"\n- 💬 Message teammate: clawteam inbox send \u003Cteam> \u003Cname> \"info...\"\n- 📨 Check inbox:      clawteam inbox receive \u003Cteam>\n- 😴 Report idle:      clawteam lifecycle idle \u003Cteam>\n```\n\nThis means **any CLI agent** can participate in a ClawTeam team — it just needs to run shell commands. No custom SDK, no API integration, no framework lock-in.\n\n---\n\n## 📖 Command Reference\n\n\u003Cdetails open>\n\u003Csummary>\u003Ch3>🔧 Core Commands\u003C\u002Fh3>\u003C\u002Fsummary>\n\n```bash\n# 🏗️ Team lifecycle\nclawteam team spawn-team \u003Cteam> -d \"description\" -n \u003Cleader>\nclawteam team discover                    # List all teams\nclawteam team status \u003Cteam>               # Show members\nclawteam team cleanup \u003Cteam> --force      # Delete team\n\n# 🚀 Spawn agents\nclawteam spawn --team \u003Cteam> --agent-name \u003Cname> --task \"do this\"\nclawteam spawn tmux codex --team \u003Cteam> --agent-name \u003Cname> --task \"do this\"\n\n# 📋 Task management\nclawteam task create \u003Cteam> \"subject\" -o \u003Cowner> --blocked-by \u003Cid1>,\u003Cid2>\nclawteam task update \u003Cteam> \u003Cid> --status completed   # auto-unblocks dependents\nclawteam task list \u003Cteam> --status blocked --owner worker1\nclawteam task wait \u003Cteam> --timeout 300\n\n# 💬 Messaging\nclawteam inbox send \u003Cteam> \u003Cto> \"message\"\nclawteam inbox broadcast \u003Cteam> \"message\"\nclawteam inbox receive \u003Cteam>             # consume messages\nclawteam inbox peek \u003Cteam>                # read without consuming\n\n# 📊 Monitoring\nclawteam board show \u003Cteam>                # terminal kanban\nclawteam board live \u003Cteam> --interval 3   # auto-refresh\nclawteam board attach \u003Cteam>              # tiled tmux view\nclawteam board serve --port 8080          # web UI\n```\n\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Ch3>🌳 Workspace, 📝 Plan, 🔄 Lifecycle, ⚙️ Config\u003C\u002Fh3>\u003C\u002Fsummary>\n\n```bash\n# 🌳 Workspace (git worktree management)\nclawteam workspace list \u003Cteam>\nclawteam workspace checkpoint \u003Cteam> \u003Cagent>    # auto-commit\nclawteam workspace merge \u003Cteam> \u003Cagent>         # merge back to main\nclawteam workspace cleanup \u003Cteam> \u003Cagent>       # remove worktree\n\n# 📝 Plan approval\nclawteam plan submit \u003Cteam> \u003Cagent> \"plan\" --summary \"TL;DR\"\nclawteam plan approve \u003Cteam> \u003Cplan-id> \u003Cagent> --feedback \"LGTM\"\nclawteam plan reject \u003Cteam> \u003Cplan-id> \u003Cagent> --feedback \"Revise X\"\n\n# 🔄 Lifecycle\nclawteam lifecycle request-shutdown \u003Cteam> \u003Cagent> --reason \"done\"\nclawteam lifecycle approve-shutdown \u003Cteam> \u003Crequest-id> \u003Cagent>\nclawteam lifecycle idle \u003Cteam>\n\n# 🎪 Templates\nclawteam launch \u003Ctemplate> --team \u003Cname> --goal \"Build X\"\nclawteam template list\n\n# ⚙️ Config\nclawteam config show\nclawteam config set transport p2p\nclawteam config health\n```\n\n| Setting | Env Var | Default | Description |\n|---------|---------|---------|-------------|\n| `data_dir` | `CLAWTEAM_DATA_DIR` | `~\u002F.clawteam` | Data directory |\n| `transport` | `CLAWTEAM_TRANSPORT` | `file` | `file` or `p2p` |\n| `workspace` | `CLAWTEAM_WORKSPACE` | `auto` | `auto` \u002F `always` \u002F `never` |\n| `default_backend` | `CLAWTEAM_DEFAULT_BACKEND` | `tmux` | `tmux` or `subprocess` |\n| `skip_permissions` | `CLAWTEAM_SKIP_PERMISSIONS` | `true` | Auto-approve agent tools |\n\nLegacy aliases:\n`OH_*` env vars are still accepted for compatibility, but new docs and examples use `CLAWTEAM_*`.\n\n\u003C\u002Fdetails>\n\n---\n\n## 🏗️ Architecture\n\n```\n  Human: \"Optimize this LLM\"\n         │\n         ▼\n  ┌──────────────┐     clawteam spawn     ┌──────────────┐\n  │ 🦞 Leader    │ ──────────────────────► │ 🤖 Worker    │\n  │ (Claude Code)│ ──────┐                │ (Claude Code)│\n  │              │       │                │ git worktree │\n  │ Uses:        │       │                │ tmux window  │\n  │ • spawn      │       │ clawteam spawn └──────────────┘\n  │ • task create│       │\n  │ • inbox send │       ▼                ┌──────────────┐\n  │ • board show │ ──────────────────────► │ 🤖 Worker    │\n  │ • task wait  │       │                │ (Codex)      │\n  └──────────────┘       │                │ git worktree │\n                         │                │ tmux window  │\n                         │ clawteam spawn └──────────────┘\n                         ▼\n                   ┌──────────────┐\n                   │ 🤖 Worker    │    Each worker uses:\n                   │ (any CLI)    │    • task list (check work)\n                   │ git worktree │    • task update (report done)\n                   │ tmux window  │    • inbox send (message leader)\n                   └──────────────┘    • inbox receive (get instructions)\n                         │\n                         ▼\n              ┌─────────────────────┐\n              │    ~\u002F.clawteam\u002F     │\n              │ ├── teams\u002F   (who) │\n              │ ├── tasks\u002F   (what)│\n              │ ├── inboxes\u002F (talk)│\n              │ └── workspaces\u002F    │\n              │     (isolated code)│\n              └─────────────────────┘\n```\n\nAll state lives in `~\u002F.clawteam\u002F` as JSON files. No database, no server, no cloud. Atomic `tmp + rename` writes ensure crash safety.\n\n| Spawn Default | Value | Override |\n|---------------|-------|----------|\n| Backend | `tmux` | `clawteam spawn subprocess ...` |\n| Command | `claude` | `clawteam spawn tmux codex ...` |\n| Workspace | `auto` (git worktree) | `--no-workspace` |\n| Permissions | skip | `--no-skip-permissions` |\n\n| Transport | How it works | When to use |\n|-----------|-------------|-------------|\n| **file** (default) | JSON files in inbox dirs | Single machine, shared FS |\n| **p2p** | ZeroMQ PUSH\u002FPULL + file fallback | Low-latency, auto-fallback |\n\n---\n\n## 🗺️ Roadmap\n\n| Phase | Version | What | Status |\n|-------|---------|------|--------|\n| **Current** | v0.3 | File + P2P (ZeroMQ) transport, Web UI, multi-user, team templates | ✅ Shipped |\n| **Phase 1** | v0.4 | Redis Transport — cross-machine messaging | 🔜 Planned |\n| **Phase 2** | v0.5 | Shared State Layer — team config and tasks across machines | 🔜 Planned |\n| **Phase 3** | v0.6 | Agent Marketplace — discover and reuse community agent templates | 💡 Exploring |\n| **Phase 4** | v0.7 | Adaptive Scheduling — dynamically reassign tasks based on agent performance | 💡 Exploring |\n| **Phase 5** | v1.0 | Production-grade — auth, permissions, audit logs | 💡 Exploring |\n\n### Milestones\n\n| Milestone | Status | Notes |\n|---|---|---|\n| v0.1.x | ✅ Shipped | Core CLI, team\u002Ftask\u002Finbox flows, board, templates, packaging. |\n| v0.2.0 | ✅ Released | Stabilization, docs refresh, spawn\u002Fworkspace fixes, and release packaging. |\n| v0.3 | 📍 Roadmap baseline | File + P2P transport, Web UI, multi-user workflow, team templates. |\n\nFor the longer-term plan, see [ROADMAP.md](ROADMAP.md).\n\n---\n\n## 🤝 Contributing\n\nWe welcome contributions! ClawTeam is designed to be extensible:\n\n- 🤖 **New agent integrations** — Add support for more AI coding agents\n- 🎪 **Team templates** — Create TOML templates for new domains (DevOps, data science, etc.)\n- 🔌 **Transport backends** — Redis, NATS, or other message transports\n- 📊 **Dashboard improvements** — Enhanced Web UI, Grafana integration\n- 📖 **Documentation** — Tutorials, best practices, agent prompt engineering guides\n\n---\n\n## 📖 Acknowledgements\n\n- [@karpathy\u002Fautoresearch](https:\u002F\u002Fgithub.com\u002Fkarpathy\u002Fautoresearch) — the autonomous ML research framework used in our 8-agent swarm demo\n- [Claude Code](https:\u002F\u002Fclaude.ai\u002Fclaude-code) and [Codex](https:\u002F\u002Fopenai.com\u002Fcodex) — AI coding agents that work as ClawTeam team members\n- [ai-hedge-fund](https:\u002F\u002Fgithub.com\u002Fvirattt\u002Fai-hedge-fund) — inspiration for the multi-analyst hedge fund template\n- [CLI-Anything](https:\u002F\u002Fgithub.com\u002FHKUDS\u002FCLI-Anything) — sister project making all software agent-native\n\n---\n\n## ⭐ Star History\n\nIf you find ClawTeam helpful, please consider giving us a star! ⭐\n\n\u003Cdiv align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fstar-history.com\u002F#HKUDS\u002FClawTeam&Date\">\n    \u003Cpicture>\n      \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Fapi.star-history.com\u002Fsvg?repos=HKUDS\u002FClawTeam&type=Date&theme=dark\" \u002F>\n      \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Fapi.star-history.com\u002Fsvg?repos=HKUDS\u002FClawTeam&type=Date\" \u002F>\n      \u003Cimg alt=\"Star History Chart\" src=\"https:\u002F\u002Fapi.star-history.com\u002Fsvg?repos=HKUDS\u002FClawTeam&type=Date\" \u002F>\n    \u003C\u002Fpicture>\n  \u003C\u002Fa>\n\u003C\u002Fdiv>\n\n## 📄 License\n\nMIT License — free to use, modify, and distribute.\n\n---\n\n\u003Cdiv align=\"center\">\n\n**ClawTeam** — *Agent Swarm Intelligence.* 🦞\n\n\u003Csub>8 agents × 8 H100s × 2430 experiments × one CLI × one swarm\u003C\u002Fsub>\n\n\u003Cbr>\n\n\u003Cimg src=\"assets\u002Ficon.png\" alt=\"ClawTeam\" width=\"80\">\n\n\u003C\u002Fdiv>\n\n\u003Cp align=\"center\">\n  \u003Cem>Thanks for visiting ✨ ClawTeam!\u003C\u002Fem>\u003Cbr>\u003Cbr>\n  \u003Cimg src=\"https:\u002F\u002Fvisitor-badge.laobi.icu\u002Fbadge?page_id=HKUDS.ClawTeam&style=for-the-badge&color=00d4ff\" alt=\"Views\">\n\u003C\u002Fp>\n","ClawTeam 是一个基于人工智能代理群智能的自动化工具，通过一条命令实现全流程自动化。其核心功能包括AI代理团队的自组织、任务分配与结果交付，支持多种AI代理如Claude Code、Codex等，并采用Python编写，兼容性广泛。该项目利用了P2P传输技术以及提供Web UI界面来增强用户体验。特别适用于需要大规模自动化的机器学习实验场景、AI模型训练与优化等领域，能够显著提高研究效率和成果产出速度。",2,"2026-06-11 03:48:30","high_star"]