Two days after Anthropic shipped Claude Fable 5.1, fable51-worlds (PhiloLabs) hit the HN front page with 248 points. It’s an open-source agent project: a swarm of fully autonomous Fable 5.1 agents researches a real place, models it, quality-checks itself, and ships a walkable Three.js world that runs in any browser. No game engine, no human 3D artist.
What the swarm actually built
The first demo is San Francisco’s Union Square: 453 OSM building footprints, 75 hand-authored façades, 129 named storefronts, 220 pedestrians walking a 1,398-node nav graph, 109 vehicles including the Powell St cable cars, plus Apple and Nintendo store interiors you can enter. Real terrain from USGS elevation, real street grid from OpenStreetMap — every fact tagged with a source and confidence level.
Why it matters
The entire build came from a single prompt, kept verbatim in PROMPT.md. Research agents ran in parallel, reviewer agents graded the output against photographs. It landed the same day World Labs launched Atlas, turning “world models” into a two-way fight: one side generates worlds with neural nets, the other has coding agents assemble them from open data. The second approach is auditable. That’s the interesting part.
You Might Also Like
- World Labs Marble fei fei li Just Raised 1 Billion to Build 3d Worlds From a Single Photo
- From Claude Flow to Ruflo 22k Stars 5900 Commits and the Multi Agent Swarm Taking Over Claude Code
- Atlas by World Labs fei fei li Turns one Photo Into a Controllable 3d World and Ships Zero Papers
- 98 of b2b Visitors Never get a Demo Naoma ai Demo Agent Raised 440k to Change That
- 27 Agents 109 Skills 88k Github Stars is Everything Claude Code Genius or Over Engineering
Top AI Product 
Leave a comment