AI 与推理 · AI & Inference

Fastest, smallest, and fully autonomous AI assistant infrastructure written in Zig

Any model. Any hardware. Zero compromise. Built with @ziglang / @openxla / MLIR / @bazelbuild

x64dbg-MCP Server is a native MCP (Model Context Protocol) plugin for x64dbg that exposes the debugger's full functionality over HTTP. Connect any MCP-compatible AI assistant and control x64dbg programmatically: set breakpoints, step through code, read memory, dump registers, and more. Built with Zig — zero dependencies, single-binary output, cros

Unix like coding agent

LLM Inference Engine for Metal, CUDA and Vulkan.

Zig code intelligence server and MCP toolset for AI agents. Fast tree, outline, symbol, search, read, edit, deps, snapshot, and remote GitHub repo queries.

Native LLM inference server for Apple Silicon. OpenAI + Anthropic API compatible. No Python. Zig backend, Swift frontend macOS app with chat, music, voice, video generation.

Management console for the Null ecosystem — install, configure, and monitor AI agents, orchestration workflows, task pipelines, and system health

Zig INferenCe Engine — Local LLM inference on AMD GPUs and Apple Silicon

A deep learning framework built on an autograd engine with high level abstractions and low level control.

Open-source AI coding desktop app with parallel agents and persistent project memory. Local models, Ollama or cloud AI. Windows, macOS and Linux.

UCI Chess Engine written in Zig.

graff — a fast agentic coding harness in Zig: multi-provider, MCP, workflows, DGM evolution loop, TS/Python SDKs

Sushi engine runs very high quality quant model in Apple Silicon, using both STQ and EXL3 format with special tuning technology.

Orchestrates multi-step AI agent workflows — defines execution graphs, manages shared state, dispatches work to agents, and tracks progress with checkpoints and streaming

Fast computer automation CLI for AI agents. Control any desktop with screenshots, clicks, typing, scrolling, and more.

CLI/GUI tool for efficient and easy safetensors and gguf model conversion

A GNU screen style terminal multiplexer built on libghostty.

DeepSeek-V4.1-Flash on 2x NVIDIA DGX Spark with the TensorFold engine: 1.8-1.9x code, 2.5x multi-stream, 1.7-1.95x prefill vs the vLLM kit, exact speculative decoding. Work in progress.

Read what your agent wrote before you say LGTM. A fast TUI in Zig for reviewing code with vim motions.

GPU accelerated terminal for agentic workflows

A Model Context Protocol (MCP) library for the Zig ecosystem.

SQL queries for CSV files. The analytical CSV query engine for AI agents cli: csvql.

An educational transformer-based LLM in pure Zig

Open-source AI agent CLI built in Zig. A pure Zig rewrite of OpenClaw — one ~5MB binary, zero Node.js. Agentic loop, SSE streaming, tool dispatch, SQLite history, and multi-backend support (Claude, Ollama, OpenAI-compatible). Cross-compiles Mac, Linux, and Windows in one command.

Big models. Small Macs. Zero excuses.

Manages tasks and persistent storage for AI agents — assigns work, tracks completion, and provides a shared key-value store with full-text search across agent sessions

A minimal coding harness that rewrites itself in Lua

Local LLMs that run lean, load instantly and keep getting smarter. A from-scratch runtime + model format in Zig: zero-copy mmap, CPU/CUDA/Vulkan engines, OpenAI-compatible server, 2:4 sparsity, Net2Net growth, in-file version history, self-learning loops.

CLI proxy that reduces LLM token consumption by 78%+. Single Zig binary under 260KB. Zero dependencies.

The Trinity ternary compute stack — tri CLI · BitNet LLM · VSA · GF16

MemXT is a Local long-term memory for Claude Code & coding agents. 100% local

A fast, lightweight TUI coding agent — a single static binary, no runtime

Drop-in replacement for `claude -p` that drives the interactive Claude Code TUI inside an in-process zmux PTY session.

A fast and compact CPU-only inference engine for Laya in Zig.

High-performance, Hardware-First Vector Database engine built in Zig and Go. Sub-millisecond HNSW search with 64-bit metadata filtering and 1.8MB binary footprint.

CPU-first, eager-execution tensor/autograd runtime and LLM inference engine written in pure Zig

Fetch any URL to clean, LLM-ready markdown. 3-tier engine: HTTP → Lightpanda → CloakBrowser.

make your ai track the progress of your project without drifting off

blazigly fast gRPC/MCP client & server implementation in zig

Linear Algebra and Deep Learning Zig library with GPGPU and heterogeneous computing

Minimal pi coding-agent re-implementation in Zig

High-performance FlashAttention-2 for AMD, Intel, and Apple GPUs. Drop-in replacement for PyTorch SDPA. Triton backend for ROCm (MI300X, RDNA3), Vulkan backend for consumer GPUs. No CUDA required.

Learn how LLMs work by building one in Zig -- from tensors to text generation.

A tiny, self-contained LLM coding agent. One binary, 180+ providers, zero runtime dependencies.

Zig GGUF inference engine for Apple Silicon

Tensor library for machine learning, inspired by ggml

Run autonomous AI agents across machines using a distributed filesystem.

A lightweight AI agent for legacy hardware. Written in Zig, runs on Windows XP SP3 x86 with 64 MB RAM. Connects to Ollama for LLM-powered system diagnostics and management via function calling.

Talu is a single-binary, local-first LLM runtime with a Zig core and multi-language bindings — CLI, Python API, HTTP server, plugin-extensible Web UI, structured output, quantization, embeddings, and unified local/remote model routing.

Zero-trust capability proxy for AI agents. Scoped access to files, git, and tools via signed tokens. No mounts, no credentials, fully audited.

moondream in zig.

zCUDA: 🔗 CUDA Binding Layer | ⚡ CUDA Kernel DSL. Write CUDA kernels in pure Zig. Or keep your C++ kernels. Ship them all to GPU.

Ollama Zig library

An LLM code exploration token optimizer that runs in a Hyperdimensional Compute MCP server

Pocket Flow: A minimalist LLM framework. Let Agents build Agents!

A lightweight, high-performance LLM SDK & Edge Router written in Zig

Zig tensor library

Zig AI Coding Assistant — multi-provider LLM agent CLI with tool calling

Competitive Pokémon battling AI

LLM tokenizer in Zig