Repos

Open-source project templatesgrokbot.com: TemplatesGrokBot.com is a library of 3,000+ ready-made Grok Bot templates. Search by task or job title, install any bot . Extends or complements Grok Bot for builders.

Gallery of games and films made by Claude Opus 5.5 and Astra, sorted by how quickly you can try each one, for side-by-side comparison.

Exposes any agent worker, Muse included, as an OpenAI-style endpoint so 9Router or other clients can use it as a custom provider.

Practical recipes for putting a fast, cheap gate model in front of costly Muse steps like browsing and deep research.

Nik Anand's Python CLI covers chat, side chats, the feed, goals, ideas and sessions, so no browser is needed to work with Muse.

Open-source Claude Code mod by whyashthakker. Its repo describes it as: Collection of 50 mods you can use with Claude Code. Shows what the community built in the first days of mods.

A tab showing your MUSE AV asset library and recent generations per workspace; click an asset to copy its link as a reference image.

Runs the typed-decisions checkpoint on Metal Performance Shaders: ~32 ms median and ~2.1 GiB RAM on M5 Pro, plus a slower ~0.74 GiB low-memory mode. Demo and benchmark included.

Packaged run recipe for a single AMD APU.

Open-source project omabot: Grok Bot in the Omarchy bar. Extends or complements Grok Bot for builders.

Animated faces under the Omarchy bar or MacBook notch show whether agents are thinking, working or waiting for approval.

Two ways to wire a Hermes agent into the Grok Bot app as a connector, via SSH relay or a native API.

Tested llama.cpp GGUF route for a single Spark or Asus GX10 box.

VLLM patch and guide for running Flash-Next on a CMP 170HX with streamed embeddings.

Built on the Gadget SDK protocol: streaming chat, VM lookup, token refresh and macOS Bluetooth pairing, so you can talk to Muse from your own Node code.

A native Android animation with shader effects drawn live, no keyframes.

Serving the model through ExLlamaV3 quantisation.

A Mario Kart-style racer in three.js with no build step and no external assets.

Convenience script that boots the model on a single DGX Spark or variant.

Installs Claude Code on the Muse agent VM and exposes it in a browser so you can code from a phone, plus a multi-model workspace called dimensio.

Community vLLM build that runs the model on cheap ex-mining SM80 hardware.

Cheap dual-card build that also scales to 800 tok/s at 23 concurrent users.

Terminal harness and evidence that Astra wrote and used while ascending NetHack 3.6.7 through Python, released so others can inspect how the run worked.