Top Picks

SmolSignal is an AI signal copilot for Flipper Zero files. It explains common capture formats, classifies safety risk, supports local and cloud model backends, and can generate safe consumer IR files.

An MCP server that gives coding agents a compiler-backed graph of a TypeScript codebase, so they can answer structure questions with fewer tokens and exact file-and-line references.

A from-scratch ~1B parameter language model on Hugging Face with open weights, open data, training details, and loading guidance. The page is unusually candid about capabilities, limits, and evaluation.

Busabase is an open-source, local-first database and knowledge base built around approval workflows for AI-generated changes. It lets users review, approve, and merge agent-proposed records, and includes a simple API plus desktop, Docker, and source-based setup options.

Kernel-level read gate for AI coding agents that blocks protected file reads before bytes reach the agent, with Linux fanotify and macOS Endpoint Security support.

A TypeScript library for deterministic context folding in long-running LLM agent sessions, designed to preserve prompts, avoid extra summarization calls, and keep provider caches hot.

Open-source AI SQL IDE for databases with a built-in AI data analyst, schema browser, notebook-based query runner, and persistent workspace state.

A deterministic Python project for empirical law discovery in elementary cellular automata, with simulation, discovery, reproducibility guides, and published preprints.

Open-source N64 game engine and visual editor built with libdragon and tiny3d. It targets real hardware and accurate emulators, with GLTF import, HDR+bloom, big-texture support, scene management, audio, collision, and node-graph scripting.

An AI-powered pipeline that turns diarized podcasts or interviews into YouTube-ready videos with semantic B-roll, burned-in subtitles, optional TTS voice replacement, and YouTube publishing.

Open-source SDK and libraries for building local-first, peer-to-peer AI apps. It supports local LLM inference, embeddings, translation, transcription, TTS, OCR, image generation, and an OpenAI-compatible API across desktop and mobile platforms.

A Python library that transpiles stateful imperative LLM workflows into optimizable DSPy programs, with typed outputs, self-correction loops, validation reporting, and prompt tuning support.

Tecture helps coding agents write and maintain architecture docs directly in the repo, then lets teams explore them as interactive diagrams in VS Code or the browser.

An open-source MCP server that returns page-cited, verbatim evidence from embedded and firmware reference docs for coding agents.

GreyFox Community Edition is a self-hosted AI traffic proxy and admin console for teams that want to control LLM usage, enforce per-user quotas, reuse cached responses, and keep traffic inside their own infrastructure.

Recall is a local-first memory tool for Claude Code that logs sessions and generates a compact project summary offline, without sending data to an API.

Open-source authorization engine for AI agents that adds confidence-based gating, human review, policy-as-code, and audit logging. The repo shows quickstart code, local demo steps, SDK installs, and self-hosting options.

Rust-based PostgreSQL MCP server with 135 tools for Claude Desktop and other MCP-compatible AI tools. Supports stdio, TCP, and HTTP/2 transports, plus quick-start install and setup instructions.

CrankGPT is a fully offline voice assistant built on a hand-crank-powered Raspberry Pi setup. The page explains the hardware stack, local speech recognition, on-device LLMs, text-to-speech, power smoothing, and latency measurements for running the system without cloud dependency.

NILScript is an open standard and CLI toolkit for letting AI agents act on real systems through a gated propose → approve → commit → rollback flow. The page shows a live playground, quickstart commands, and benchmark claims around unauthorized writes.

An MCP server for the PostFast API that lets AI tools schedule, manage, and analyze social media posts across major platforms.

MakerChecker is a self-hosted governance layer for AI agents that enforces roles, approvals, and tamper-evident audit logs. The repo shows quickstart commands, a demo flow, and SDK/connectors for TypeScript, Python, and LangChain-style integrations.

ctx is a local-first desktop workbench for coding agents, now open source under GPL-3.0. The page explains that it runs agents in a desktop environment with worktrees, transcripts, artifacts, diff review, container boundaries, and a local merge queue.

A macOS menu bar app that helps users pace Claude Code and Codex usage against weekly and 5-hour quota windows.