Explore the directory

An agent harness that compiles a model's whole forward pass into a single, provably correct, self-tuning CUDA megakernel that beats cuBLAS at batch-1 LLM decode on several GPUs.

• Coding agents
• Inference

A Kotlin DSL and scripting framework for defining LLM-powered AI agent solutions.

• Agent frameworks

A heterogeneous prefill/decode inference bridge for DeepSeek-V4-Flash that runs prefill on NVIDIA GPUs with vLLM and decode on Apple Silicon with oMLX, connected over ordinary networking.

• Inference

A Python implementation of the PALI-3 vision-language model, which encodes images with a contrastively pretrained SigLIP vision transformer and generates text through a UL2 encoder-decoder transformer.

• Models

An AI coding agent with a single Python core and three interchangeable front ends, a headless CLI, a Textual TUI, and an Electron desktop app, that speaks any OpenAI-compatible API.

• Coding agents

A terminal coding agent that sends every change to two independent models from different providers for review, shipping a change only when both approve.

• Coding agents
• Code quality

An offline MCP server that turns an AI assistant's description of a system into a laid-out, hand-drawn Excalidraw diagram, covering 25 diagram types from architecture to Gantt charts.

• MCP servers
• Design tools

A hands-on book and Go codebase that teaches harness engineering by building a working terminal AI coding agent from scratch and experimenting with its providers, tools and permissions.

• Coding agents
• Books

An MCP server for Blender that exposes over 180 tools for modelling, materials, shading, animation and rendering, letting Claude or any MCP client control Blender through natural language.

• MCP servers
• Design tools

A curated list of projects, libraries and articles for building LLM-powered Rust applications with the Rig framework.

• Agent frameworks
• Curated lists

An open, benchmark-tested context format that stores a codebase's AI-agent rules and a per-file boundary index in a centralized .aigx directory without injecting anything into source files.

• Agent skills

A production-oriented Next.js starter template for rebuilding a live website as a typed Next.js application using AI coding agents, with a reproducible multi-step clone workflow.

• Coding agents
• Web frameworks

A syntax dialect that adds JSX-style expressions to OCaml, compiling them into plain OCaml function calls through a preprocessor.

• Languages

A discrete-event simulator for modern LLM serving systems, used to study parallelism, runtime optimizations and disaggregated serving without deploying to real GPU clusters.

• Inference

A local-first memory tool for coding agents that stores decisions, bug fixes and project context as markdown, indexed with full-text search and optional semantic search.

• Agent memory

A local-first Streamlit application for document analysis that combines hybrid retrieval, optional knowledge-graph search and multi-agent coordination over local large language models.

• RAG

An installable catalog of thousands of AI coding skills that can be installed into CLI, API, MCP and A2A clients with a single command.

• Agent skills
• Curated lists

A terminal-native autonomous coding agent for real repositories, distributed as a self-contained binary that supports seven model providers and non-interactive use in CI.

• Coding agents

An evaluation library for African-language and code-switching NLP systems, with metrics for dialectal robustness, terminology accuracy and hallucination that mainstream benchmarks do not measure.

• AI evals

A local-first Python toolkit and FastAPI service for inspectable retrieval-augmented literature review, showing the retrieved passages and run history behind each answer.

• RAG

A community-maintained collection of markdown skills that give AI coding assistants guidance for writing correct MapLibre GL JS mapping code.

• Agent skills

A frozen-task, frozen-hardware public leaderboard for speedrunning LoRA fine-tuning of small language models, with every record independently re-verified on a cloud GPU sandbox.

• Models
• AI evals

A Rust workflow engine that orchestrates Claude Code and Codex through a plan, review-quorum, implement, review and fix pipeline, using Git worktree isolation for multiple AI coding agents.

• Coding agents

An open-source terminal AI coding agent written in Rust as a single binary, supporting more than 15 LLM providers and over 1,000 models as a self-hostable alternative to Claude Code and Codex CLI.

• Coding agents