GitHub Repos Worth Saddling Up
Stars land in Learning; watches stay on the radar. Club projects and friends are curated by hand.
Stuff I’m Learning From
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
★ 191,647 starsA Claude Code skill/plugin (and 30+ other agents) that makes the agent talk like a caveman—cuts ~75% of output tokens while keeping full technical accuracy. Brain still big. Mouth small.
★ 120,547 starsFilters and compresses command outputs before they reach your LLM context—60–90% fewer tokens on common dev commands via a single Rust binary.
★ 86,698 starsCompresses everything your AI agent reads—tool outputs, logs, RAG chunks, files, and conversation history—before it reaches the LLM. Same answers, fraction of the tokens.
★ 79,776 starsA hierarchical inference engine for frontier MoE models: hot experts on the GPU (CUDA, Vulkan), warm ones in RAM, cold ones streamed from NVMe. Written in C, runs on the hardware you already own.
★ 51,447 starsReverse Engineering / Authorized Penetration Testing / Security Research Skill Router Pack AI-powered routing + On-demand toolchain bootstrapping + Self-evolving knowledge base Supports Claude Code, Kiro, Cursor, Cline, and other AI coding clients 逆向/渗透/安全技能路由包 - AI 自动路由 + 按需自举工具链 + 自动进化经验库 | 支持 Claude Code / Kiro / Cursor / Cline 等代码 AI 客户端
★ 44,737 starsMultiplayer agent harness for teams
★ 15,712 starsGoogle’s open-source distributed agent runtime for building and running multi-agent systems at scale.
★ 14,732 starsEnterprise AI platform with guardrails, MCP registry, gateway, and orchestrator for deploying and governing agent workflows.
★ 4,445 starsSwitchyard lets LLM applications route traffic across models and providers while preserving native OpenAI and Anthropic API compatibility - enabling flexible model selection, benchmarking, and cost/performance optimization.
★ 3,560 starsAn extensible, state-of-the-art framework for columnar compression, and the fastest FOSS columnar file format. Formerly at @spiraldb, now an Incubation Stage project at LFAI&Data, part of the Linux Foundation.
★ 3,320 starsAn embedded local, agent-native graph database supporting Cypher 25 written in Rust. Memory or transactional filesystem.
★ 283 starsAI-powered multi-agent system for NetDevOps and security automation with human-in-the-loop workflows.
★ 14 starsIntelligent persistent memory system for LLM agents using Firebolt Core with HNSW vector search—upstream inspiration for LAML.
★ 2 starsMy Projects
Smart SunSynk inverter control tuned for the Intelligent Octopus Go tariff, focusing on house-battery optimisation and grid-friendly charging.
★ 2 starsLocal Agent Memory Layer (LAML): an intelligent persistent memory system for LLM agents using a local vector DB and HNSW search.
★ 1 starsPLG IDE & Loveable app starter template: turn your database or platform into an interactive, product-led demo experience in the IDE and on the web.
★ 0 starsSelf-hosted local server to keep PLAATO Keg devices working after the cloud shutdown—open-plaato-keg, MQTT, and Docker Compose.
★ 0 starsOn my radar
Training-free language model modification: Remove censorship, suppress slop, reduce sycophancy, and more
★ 40,714 starsOpen-source vacuum robot cleaner
★ 13,106 starsGPU-native composable analytics engine
★ 1,109 starsRun any agent on any model, models stored as OCI images
★ 590 starsClaude Code plugins for the work that keeps coming back: repo maps, conditional rules, ticketed parallel work, extra subscription models, local usage metrics, and guided setup.
★ 285 starsFalkorDB-MCPServer is an MCP (Model Context Protocol) server that connects LLMs to FalkorDB
★ 42 starsCARS — Claude Agent Registration Server: an experimental peer-session discovery ledger for Claude Code (MCP server + hooks + skills, installable plugin)
★ 1 starsA modest Go port of JTON (JSON Tabular Object Notation / Zen Grid) by Gowthamkumar Nandakishore (arXiv:2604.05865) — byte-for-byte conformant with the reference impl.
★ 0 stars