Explore the directory

A scenario-based benchmark for local LLM inference on Apple Silicon that measures effective end-to-end throughput, including the prefill phase, across real workloads such as agent conversations and document classification.

• Inference
• AI evals

A Nix script that runs coding agents such as Claude Code, Codex CLI and GitHub Copilot CLI inside hardware-isolated QEMU microVMs, so their dangerous auto-approve flags cannot touch the rest of the host.

• Coding agents
• Containers

A free, open-source local LLM server that turns an idle Mac or GPU into a public, OpenAI-compatible inference node reachable over HTTPS, sharing spare compute with friends, apps or AI agents.

• Inference

A VL-JEPA-inspired pipeline that compresses images, text, conversations and RAG documents locally through Ollama before sending only the compact result to a cloud LLM API, cutting token costs by around 80 percent.

• Inference

A local-first CUDA image-generation app built around the Rayzist Z-Image Turbo fine-tune, with a FastAPI backend, browser UI and CLI, LoRA and wildcard libraries, prompt enhancement and upscaling.

• Generative media

A terminal emulator for Mac with built-in local AI that explains failed commands, suggests what to run next, and answers questions about the current session, entirely offline through llama.cpp with no API key.

• Terminals

A fork of the HumanEval evaluation harness adapted to benchmark how well large language models generate correct Python code for bio-image analysis tasks.

• AI evals

An experimental agent orchestration loop built on the Pi coding agent that takes a work item through implementation specialists and then independently reviews the resulting local pull request through separate PR-control specialists.

• Coding agents

An open-source TypeScript agent runtime and SDK for durable conversations, tool and MCP execution, approvals and traceable activity, letting a product keep its own architecture while adopting only the layers it needs.

• Agent frameworks

A programming language and runtime for building AI agents that manages model calls, tool execution, permissions and history, so the same agent code can run against local or hosted models.

• Agent frameworks
• Languages

An MCP server that runs inside Home Assistant, letting AI assistants control smart-home devices, query states, run services and manage automations through natural language.

• MCP servers

An MCP framework giving AI assistants unified access to Gmail, Drive, Docs, Sheets, Slides, Calendar, Forms, Chat, Photos and Contacts, with a code-mode execution sandbox and Qdrant-backed semantic memory of past tool calls.

• MCP servers

A hands-on Microsoft Foundry hackathon lab that walks developers through designing, monitoring, evaluating and orchestrating enterprise AI agents with the Foundry SDK and portal.

• Tutorials

A hackable, open-source terminal coding agent and agent runtime with a browser and mobile web dashboard for managing every session, model switching, subagents, and a built-in issue-to-merge automation workflow.

• Coding agents

A single-file, dependency-free transparent proxy that lets the Codex CLI or desktop app call NVIDIA NIM's free-hosted models, such as DeepSeek V4 Pro and Qwen3 Coder, using a user's own NVIDIA account.

• LLM gateways

A self-hosted, open-source multi-agent AI orchestration platform that runs each agent in its own isolated Docker container, with RBAC, token budgets, audit logs and support for several messaging channels.

• Agent frameworks
• Containers

A minimal, zero-dependency LLM inference engine in pure C11 for GGUF models, with NEON and AVX2 SIMD CPU backends, mixture-of-experts routing, TurboQuant KV-cache compression, and a WebAssembly build.

• Inference

A curated, community-sourced collection of prompts for OpenAI's GPT Image 1.5 model, with images for each prompt and a companion web gallery offering search and one-click generation.

• Generative media
• Curated lists

A structured, markdown-based synthetic dataset covering business, technology, psychology and other topics, created to give AI training, testing and question-answering research a consistent, well-organized knowledge source.

• Models

Integrations connecting AgentQL, a tool for extracting structured data from web pages by query or natural-language prompt and interacting with pages through Playwright, to workflow-automation and AI agent frameworks.

• Workflow automation
• Browser automation

An open-source AI coding agent for the terminal that remembers what it learns about a user and their projects from the work itself, avoiding re-explaining context on every session.

• Coding agents

A patched, native-Windows build of vLLM adding three Windows-specific fixes, a Qwen3 reasoning parser and tool-calling backports, published as prebuilt wheels for both Ampere/Ada and Blackwell RTX GPUs.

• Inference

The Super Tiny Compiler, a well-known minimal educational compiler example, reimplemented in Ada, taking a small Lisp-like expression and compiling it to a line of C-style function-call code.

• Languages
• Tutorials

A native desktop app, built with Rust and Tauri, that gives a single dashboard over every local development project, with visual package management and dependency tracking across React, Angular, Next.js and Vue projects.

• Desktop utilities