ntindle/spark-vm
Setup + tooling for the spark-vm dev box
Open Source Maintainer of AutoGPT
Setup + tooling for the spark-vm dev box
Pre/post authoritative game states for phase-rs/phase open bug repros (oldest to newest).
Fork of CLIProxyAPI: routes by quota reset time, discovers Claude and Codex models live, ships an Unraid-ready image. See FORK.md.
tmux-resurrect for herdr — snapshot workspaces, tabs, panes, cwd, running programs and agents, and restore them after a crash or reboot.
The missing integration layer for AI agents. Let them call any OpenAPI / MCP / GraphQL / custom js functions in secure environment.
Unofficial Unraid templates for self-hosted AI tooling
Persistent memory for Muse Code — powered by Supermemory
Moved to ntindle/unraid-apps (archived). Unofficial Unraid template for the Executor MCP gateway
Moved to ntindle/unraid-apps (archived). Unofficial Unraid template for the ntindle fork of CLIProxyAPI
An experimental open-source attempt to make GPT-4 fully autonomous.
A Magic: The Gathering rules engine and game client — Rust + WASM + React
Open-source desktop and web app for Meta's Muse Code CLI. Runs Muse natively on Windows, macOS and WSL2, with a file viewer, inline diffs, approvals and signed auto-updating installers.
Experimental desktop builds of the AutoGPT Platform. A parallel fork of Significant-Gravitas/AutoGPT that tracks upstream dev.
Run multiple Claude Desktop accounts simultaneously. Each account gets its own permanent profile and stays logged in forever.
Keep Claude Code Desktop conversations synchronized across every account on your machine.
Hoverboard Hack Firmware Generation 2 for the Hoverboard with the two Mainboards instead of the Sensorboards (See Readme.md).
GD32 Keil code project for the Gen2.x hoverboard
with different defines_2-x.h for different board layouts :-) Compiles with Keil version 6
A fork of FreeCAD that has a clanker brain.
Parametric CAD and simulation toolkit for a four-axis flyer winder for small BLDC stators.
A collaborative book on algorithms
Examples and guides for using the OpenAI API
Stop configuring your AI stack. Start using it. One command brings a complete pre-wired LLM stack with hundreds of services to explore.
Matter (formerly Project CHIP) is creating more connections between more objects, simplifying development for manufacturers and increasing compatibility for consumers, guided by the Connectivity Standards Alliance (formerly Zigbee Alliance).
One llama.cpp flag unlocks +33-39% decode speed for Qwen3.8-27B on consumer GPUs. The MTP head already ships inside your GGUF. Recipe, paired benchmarks, probe tool.
A native macOS menu bar app that sets your desktop wallpaper to a different impressionist painting each day, displayed on museum-style mats.