Production-grade recipe for DeepSeek-V4-Flash-0731 (284B MoE) on 2x NVIDIA DGX Spark: self-healing 2-node vLLM cluster, reboot-verified, tuned DSpark speculative decoding (~75 tok/s), full benchmar…

An experiment in durability

A fast, secure terminal workspace that lives in your browser

AI inference, packed simply. A blazing-fast, zero-dependency WebGPU runtime to run GGUF models directly in the browser. Features a symmetric API for seamless local execution and cloud provider rout…

Mixture-of-experts (MoE) training megakernel for NVL72s

flow is a Go library for durable, event-driven execution backed Postgres

Use Nix and NixOS on your DGX Spark!

The original Command & Conquer: Generals Zero Hour engine, ported to WebAssembly and playable in the browser.

Hybrid Marlin+CUTLASS NVFP4 kernel plugin for vLLM on sm_120 — +95% prefill at same decode (Qwen3.6-27B, RTX PRO 6000)

Unified LLM provider library in Go across Anthropic, OpenAI, vLLM, OpenRouter and many more

Everything I know about running LLMs locally

A multi-brain LLM council for your coding agent

An incremental parsing system for programming tools

deep code review agent with configurable intelligence

⚡A CLI tool for code structural search, lint and rewriting. Written in Rust

Google Cloud Knowledge Catalog Tools and Samples

Model export recipes, Python primitives, and Swift runtime utilities for on-device AI

Skills for Real Engineers. Straight from my .agents directory.

Open-source observability tool that uses AI agents to self-heal your software

Generate draw.io diagrams from natural language — 11 presets (UML, SysML/MBSE, BPMN, network, C4…), 36 tools: codebase/CI/infra-to-diagram, image→editable diagram, mind maps, build-up animation, ex…

Xquik agent Skill for REST, MCP, webhooks, exports, monitoring, tweet search, and confirmation-gated publishing. Not affiliated with X Corp.

Build animated explainer videos with your coding agent

YouTube Transcript API skills for AI agents. Get transcripts, search videos, browse channels. Works with OpenClaw, Hermes-Agent, Claude Code, Cursor, Windsurf.

Test your prompts, agents, and RAGs. Red teaming/pentesting/vulnerability scanning for AI. Compare performance of GPT, Claude, Gemini, DeepSeek, and more. Simple declarative configs with command li…

A suite of plugins for legal workflows

Not quite vibe-coding. Not quite boomer coding. Something else.

LLM speculative inference server for consumer & heterogeneous hardware

Open-source engine that turns any website into API for AI agents.

Skill to give Claude Code (and any coding agent) the ability to generate beautiful and practical Excalidraw diagrams.

Help AI coding agents write modern Go