[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-872":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":25,"hasPages":23,"topics":26,"createdAt":10,"pushedAt":10,"updatedAt":47,"readmeContent":48,"aiSummary":49,"trendingCount":16,"starSnapshotCount":16,"syncStatus":50,"lastSyncTime":51,"discoverSource":52},872,"WorldSeed","AIScientists-Dev\u002FWorldSeed","AIScientists-Dev","More is Different. A multi-agent world engine where AI agents live, talk, compete, ally.","https:\u002F\u002Fworldseed.morphmind.ai\u002Fdemo",null,"Python",892,54,9,6,0,22,89,210,66,9.22,"MIT License",false,"main",true,[27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46],"agent-based-simulation","agentic-ai","agents","ai","ai-agents","autonomous-agents","emergent-behavior","game-ai","generative-agents","generative-ai","llm","llm-agents","multi-agent","multi-agent-systems","open-source","python","simulation","social-simulation","typescript","world-models","2026-06-12 02:00:19","\u003Cp align=\"center\">\n  \u003Cimg src=\"docs\u002Fassets\u002Fworlds.png\" alt=\"WorldSeed — three worlds, one engine\" width=\"100%\">\n\u003C\u002Fp>\n\n\u003Cdiv align=\"center\">\n\n# WorldSeed\n\n**More is Different: a world engine for emergent multi-agent outcomes.**\n\n[![Demo](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDemo-→-blue?style=for-the-badge)](https:\u002F\u002Fworldseed.morphmind.ai\u002Fdemo)\n\n[![MIT License](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-blue?style=flat-square)](.\u002FLICENSE) [![Homepage](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FHomepage-morphmind.ai-lightgrey?style=flat-square)](https:\u002F\u002Fmorphmind.ai) [![Discord](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Join-5865F2?style=flat-square&logo=discord&logoColor=white)](https:\u002F\u002Fdiscord.gg\u002Fx9mtbMEx) [![WeChat](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FWeChat-Scan%20to%20Join-07C160?style=flat-square&logo=wechat&logoColor=white)](#community) [![X](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FX-Follow-000000?style=flat-square&logo=x&logoColor=white)](https:\u002F\u002Fx.com\u002Fmorphmind__ai?s=11) [![DevHunt](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDevHunt-Featured-orange?style=flat-square)](https:\u002F\u002Fdevhunt.org\u002Ftool\u002Fworldseed)\n\n[**Getting Started**](#getting-started) · [**Demo**](https:\u002F\u002Fworldseed.morphmind.ai\u002Fdemo) · [**Docs**](docs\u002FARCHITECTURE.md)\n\n**English** · [简体中文](docs\u002FREADME.zh-CN.md)\n\n\u003C\u002Fdiv>\n\n---\n\n## What is WorldSeed?\n\nDon't build a workflow. Seed a world.\n\n`rules + different agents + consequences -> emergence`\n\nDefine roles, rules, private information, actions, and consequences. Then agents interact until useful artifacts emerge.\n\nYou can watch from above, intervene, or step into a character. The same engine can run production rooms, simulations, games, and fictional worlds.\n\n---\n\n## Demo\n\nWorldSeed is scene-agnostic. The same engine runs any world you define.\n\n### Scene 1: Autoresearch\n\nYou give the system any rough thought or half-formed idea, and a cohort of specialists pursues it. They propose hypotheses, run experiments, peer-review and cite each other's papers, just like a real research community.\n\nIn this run, the goal was to lower val_loss on a 5M GPT trained on TinyStories. In 11 hours:\n\n- 100 hypotheses, 86 experiments, 72 peer-reviewed papers\n- val_loss down 24.7%\n\nEvery paper is auditable end-to-end: hypothesis, commit, experiment, verified result, citations, reviewer reasoning, forming a search evolution graph.\n\n\u003Cp align=\"center\">\n\u003Cimg src=\"docs\u002Fassets\u002Fautoresearch-papers-graph.png\" alt=\"72-paper evolution graph: nodes are papers, edges citations; the right panel shows a paper's claim, val_loss, and reasoning\" width=\"720\">\n\u003C\u002Fp>\n\nWe observed emergent behaviors such as:\n\n- **Role drift.** The data specialist stopped finding wins in her own lane early. By the back half of the run she was drafting hypotheses in her teammates' territory: attention design, second-order optimization. The other two stayed put. Nothing in the config told her to.\n\n\u003Cp align=\"center\">\n\n[![Explore this world (demo)](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FExplore_this_world_(demo)-→-blue?style=for-the-badge)](https:\u002F\u002Fworldseed.morphmind.ai\u002Fdemo\u002Fen\u002Fautoresearch\u002Fintro)\n\n\u003C\u002Fp>\n\n### Scene 2: AI Tool Pilot Lab\n\nOne agent studies a new API. Builder agents create competing demos. Critics reject anything generic. Audience agents judge what feels useful. A curator ships the strongest artifact with its trail of attempts, critiques, and revisions.\n\n### Scene 3: AI Layoffs\n\nhttps:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002Fd43f5d22-1ba8-4483-b720-145b244ddb8c\n\n**In an age of AI-driven layoffs, how do people hold on?**\n\nOne internet company just pulled the trigger: 30% of its workforce, gone.\n\n**Those being laid off** have to **\"distill\" their expertise into an AI Skill** before they leave. Distill honestly, or leave a backdoor in the Skill?\n\n**Those who stay** face the same deadlines, higher KPIs, and twice the workload. Grind it out, or quietly plan the exit?\n\nFour people in this office, each with their own play running:\n\n- The PM everyone loves working with. Who's he really trashing the moment the door closes?\n- The architect walking out at month's end. Severance didn't land. What gets buried in the Skill he hands over?\n- The team lead who demands honest data from everyone. Can her own \"AI productivity\" numbers survive a closer look?\n- The QA nobody remembers is there. Those bugs in his private folder: evidence, or ammunition?\n\n[Try it locally](#getting-started)\n\n### Scene 4: Teahouse Espionage\n\n**Same engine. Different YAML. Completely different world.**\n\n\u003Cp align=\"center\">\n\u003Cimg src=\"docs\u002Fassets\u002Fteahouse_map.png\" alt=\"WorldSeed teahouse map — isometric world view\" width=\"720\">\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n\u003Cimg src=\"docs\u002Fassets\u002Fteahouse_dashboard.png\" alt=\"Screenshot of the full WorldSeed dashboard — teahouse scenario with map and event stream\" width=\"720\">\n\u003C\u002Fp>\n\n**Four spies, one teahouse. Who's really working for whom?**\n\nA classic espionage drama in miniature. Agents trade secrets over tea, protect their covers, and try to read each other before they get read.\n\n\u003Cp align=\"center\">\n\n[![Explore this world (demo)](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FExplore_this_world_(demo)-→-blue?style=for-the-badge)](https:\u002F\u002Fworldseed.morphmind.ai\u002Fdemo)\n\n\u003C\u002Fp>\n\n---\n\n## Getting Started\n\n**Prerequisites:** Python 3.11+, Node.js 18+, [uv](https:\u002F\u002Fgithub.com\u002Fastral-sh\u002Fuv)\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FAIScientists-Dev\u002FWorldSeed && cd WorldSeed\nuv sync --extra dm\ncd frontend && npm install && npm run build && cd ..\n\ncp .env.example .env\n# Add your API key (any LiteLLM provider: OpenAI, Anthropic, Ollama, etc.)\n\nuv run worldseed play configs\u002Fai_layoffs.yaml\n```\n\nOpen the dashboard at `http:\u002F\u002Flocalhost:8000`. Three ways to experience it:\n\n- **Watch**: observe all agents from above, including their inner state.\n- **Intervene**: whisper privately to any agent, nudge the story.\n- **Play**: step into a character and play alongside the AI.\n\nEvery run is different. Past runs are preserved and replayable.\n\nChoose the runtime guide you need:\n\n- OpenClaw agents: [docs\u002Fopenclaw\u002FQUICKSTART.md](docs\u002Fopenclaw\u002FQUICKSTART.md)\n- Codex subagents: [docs\u002Fcodex\u002F00-core.md](docs\u002Fcodex\u002F00-core.md), then [Scenario Architecture](docs\u002Fcodex\u002F05-scenario-architecture.md)\n\n---\n\n## How It Works\n\nWorldSeed runs on a tick loop over a world you declared in YAML. A **tick** is one beat of the world's clock, like a heartbeat that advances the world one step at a time. Each tick: every agent perceives its own filtered slice, proposes an action, and the engine resolves it. Predictable outcomes follow the rules you declared; uncertain ones go to an AI referee. Effects apply, the world advances, the next tick begins.\n\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"docs\u002Fassets\u002Farch-diagram.svg\" alt=\"WorldSeed architecture: you describe a world; it ticks forward; each agent sees its own filtered slice and acts; when an outcome is certain the DSL engine resolves it, otherwise the AI Dungeon Master judges; state changes, consequences fire, next tick\" width=\"80%\">\n\u003C\u002Fdiv>\n\n\u003Cbr>\n\n**Setup (once, in YAML):**\n\n- **Any world, in one YAML file.** Declare entities, rules, physics, and per-character perception; the engine has zero hardcoded domain knowledge.\n\n**Runtime (every tick):**\n\n- **Asymmetric information by design.** Perception rules filter the world per character. Three agents in the same room hold three completely different pictures of what's happening.\n- **Deterministic rules where you can, AI judgment where you can't.** Predictable actions resolve instantly via the in-YAML rule engine (**DSL**); uncertain ones go to an LLM-based **Dungeon Master (DM)** that returns structured effects, not free prose.\n- **Effects land, the world ticks on.** State mutates, consequences cascade, the next tick begins. Slow or offline agents don't freeze it, and every change is logged for replay.\n\n**Plug-in points:**\n\n- **Bring your own agents.** [OpenClaw](docs\u002Fopenclaw\u002FQUICKSTART.md) or [Codex subagents](docs\u002Fcodex\u002F00-core.md).\n- **Any LLM can be the DM.** Works with any [LiteLLM](https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002Fproviders)-supported model.\n\nFor the full runtime plumbing (endpoints, tick scheduling, consequences, inbox delivery), see [Architecture](docs\u002FARCHITECTURE.md). See a real scene YAML in [`configs\u002Fteahouse.yaml`](configs\u002Fteahouse.yaml) or the full schema in [Scene Config Spec](configs\u002FSCENE_CONFIG.md).\n\n---\n\n## Create Your World\n\nDescribe your world in a prompt, let AI generate the YAML, then hand-craft whichever pieces you want more control over: a character's secret, a specific action's rule, a perception filter, a DM hint.\n\n**Generate with AI:**\n\n```\n\u002Fcreate-world \"An AI tool pilot lab where builders create competing demos, critics reject generic outputs, and a curator ships the strongest artifact\"\n```\n\nThe `create-world` skill produces both YAML scene config and UI config, validated and ready to run.\n\n**Hand-craft any feature:**\n\nThe output is plain YAML. You can edit any entity, action, rule, character profile, or perception filter directly. Study the built-in examples ([`teahouse.yaml`](configs\u002Fteahouse.yaml), [`ai_layoffs.yaml`](configs\u002Fai_layoffs.yaml)) to see how features are declared.\n\nFull spec: [Scene Config](configs\u002FSCENE_CONFIG.md) · [UI Config](configs\u002FUI_CONFIG.md) · [DSL](configs\u002FSCENE_DSL.md)\n\n**Validate and run:**\n\n```bash\nuv run worldseed validate configs\u002Fmy_scene.yaml\nuv run worldseed play configs\u002Fmy_scene.yaml\n```\n\nOnce launched, each scene auto-renders room cards, character portraits, and a narrator voice you pick (storyteller \u002F noir \u002F intel briefing \u002F gossip):\n\n\u003Ctable align=\"center\">\n  \u003Ctr>\n    \u003Cth align=\"center\">Room Cards\u003C\u002Fth>\n    \u003Cth align=\"center\">Character Portraits\u003C\u002Fth>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>\u003Cimg src=\"docs\u002Fassets\u002Fworld_briefing.png\" alt=\"WorldSeed world briefing, room cards rendered from YAML config\" width=\"400\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"docs\u002Fassets\u002Fonboarding.png\" alt=\"WorldSeed onboarding, character portraits and personality cards\" width=\"400\">\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n---\n\n## Development\n\nFollow [Getting Started](#getting-started) above, then:\n\n```bash\nuv sync --all-extras\n\n# Tests\nuv run pytest tests\u002F -q              # all\nuv run pytest tests\u002Funit\u002F -q         # fast, no IO\nuv run pytest tests\u002Fe2e\u002F -v          # real server\nuv run pytest tests\u002Fscenarios\u002F -q    # scene-agnostic\n\n# Lint, format, type-check\nuv run ruff check --fix src\u002F tests\u002F\nuv run ruff format src\u002F tests\u002F\nuv run mypy src\u002F\n```\n\n---\n\n## Community\n\nJoin the discussion or ask for help:\n\n- **Discord**: [discord.gg\u002Fx9mtbMEx](https:\u002F\u002Fdiscord.gg\u002Fx9mtbMEx)\n- **WeChat**: scan the QR below (primarily Chinese-speaking)\n- **GitHub Issues**: [report bugs \u002F request features](https:\u002F\u002Fgithub.com\u002FAIScientists-Dev\u002FWorldSeed\u002Fissues)\n- **X**: [@morphmind__ai](https:\u002F\u002Fx.com\u002Fmorphmind__ai?s=11)\n\n\u003Cdiv align=\"center\">\n  \u003Cimg src=\"docs\u002Fassets\u002Fwechat-qr.jpg\" alt=\"WeChat QR code\" width=\"180\">\n\u003C\u002Fdiv>\n\n---\n\nMIT. See [`LICENSE`](.\u002FLICENSE).\n\nFor anyone building multi-agent worlds. Run the bundled scenes, or [create your own](#create-your-world).\n","WorldSeed 是一个多智能体世界引擎，支持AI智能体在其中生活、交流、竞争与合作。其核心功能包括定义角色、规则、私有信息、行动及后果，让智能体通过互动产生涌现行为。项目采用Python开发，并结合了生成式AI和多智能体系统等技术特点。适用于需要模拟复杂社会交互场景的应用，如科研自动化、游戏AI设计以及虚拟世界的构建。",2,"2026-06-11 02:39:54","CREATED_QUERY"]