Repos

A large three.js coastal-town scene by Dan Greenheck with a live demo, the project behind his viral token-cost video.

Large awesome-list collecting cases, demos and papers on using Astra for robotics and embodied AI, maintained by a researcher and open to pull requests.

A repo of launch demos from riba2534, including a pelican-riding-a-bicycle test that also has a hosted page.

Open-source set of dozens of film styles, each pairing a reusable style prompt with a short sample film rendered purely in Canvas/WebGL code by Opus 5.5, plus director, storyboard and technique guides.

Step-by-step recipe to squeeze the 125B model onto one DGX Spark instead of two, with config and measured throughput.

Jeremy Perret measured the official reference video and rebuilt the avatar as a single filled shape with 14 states and independent eyes, no animation library.

Bilingual catalogue of 168 reviewed Claude 5.5 video cases with sampled frames, source-post links, production-path guidance and CSV data.

Composio's Karan Vaidya launched an open-source dots that runs on your Mac with each dot holding its own logged-in browser, triggers on Gmail and Calendar plus 1,500 apps, and voice calls. Pitched at a tenth of the cost.

Each style is a film Opus 5.5 wrote in code, with the prompt and full source so you can copy the look for your own videos.

Container-based server tuned for the gfx1151 Strix Halo APU, aimed at squeezing maximum tokens per second from the 125B model on a mini-PC.

Repo shares AGENTS.md house rules, nine team playbooks, a bot-role catalogue and a log of what broke during the Grok Bot team's 72-hour live build.

Wraps Muse chat, text-to-image and text-to-video behind an OpenAI-style API with multi-account rotation and automatic 48-hour token renewal.

Open-source Claude Code mod by karanb192. Its repo describes it as: Community catalogue of public Claude Code mods (function hooks), scanned from GitHub with what each mod can read, write, run or send over the network. Browse https://mods.aidojo.si/. Shows what the community…

Research code treating a multimodal model as a few-shot robot learner, with Astra reading demonstrations in context and then acting.