从 GitHub、Hacker News、arXiv 与 Product Hunt 信号中筛出的 AI 项目。评分只展示上游数据，不在界面中虚构。
40 个项目与讨论Point an agent at any website; audit it against modern-UX principles and Modern Web Guidance, then report and fix in a hill-climb loop. Built on Chrome DevTools MCP, inspired by memory-tracer. stars: 10 · forks: 1 · watchers: 10 AI relevance: ai, agent, mcp; High-value tags: agent, mcp; GitHub traction: 10 stars, 1 forks
Long-term memory for AI apps that keeps evidence, inference, and conflicts distinct. stars: 4 · watchers: 4 AI relevance: ai, llm, agent, mcp; High-value tags: agent, mcp, inference, local; GitHub traction: 4 stars, 0 forks
Obsidian Turbocharged — governed, agent-ready Obsidian MCP server. 163 tools across 31 domains, multi-vault native, pluggable embeddings. TypeScript + Rust. AGPL-3.0-only. stars: 8 · forks: 2 · watchers: 8 AI relevance: ai, agent, mcp; High-value tags: agent, mcp; GitHub traction: 8 stars, 2 forks
Which properties of interactive agent experience predict post-training transfer to unseen tasks, tools and environments? Measurement study: data-pool audit, training/eval loop, evaluation sensitivity analysis. AI relevance: ai, llm, agent, eval; High-value tags: agent, eval; Freshness: 0.0h old
A cognition aware database engine for AI agent memory. Purpose built in Rust with WAL, HNSW, knowledge graphs, and speculative context pre assembly. Not a wrapper, a ground up storage engine that thinks. stars: 118 · forks: 8 · watchers: 118 AI relevance: ai, llm, agent, rag; High-value tags: agent, rag; GitHub traction: 118 stars, 8 forks
Free proxy list, verified around the clock: HTTP, HTTPS, SOCKS4 and SOCKS5 proxies with country, anonymity, latency and uptime on every row, updated several times a day. Live list, proxy checker, keyless API and MCP server at hproxy.com. stars: 73 · forks: 6 · watchers: 73 AI relevance: ai, agent, mcp; High-value tags: agent, mcp; GitHub traction: 73 stars, 6 forks
CLI + MCP for agent context: a task layer over TickTick, Notion, GitHub, Obsidian & more (RRF retrieval, provenance, reviewed writes, undo) plus a knowledge-graph layer — ats kg: facts from any source, human-ratified. Recommended engines: Graphiti (server), LadybugDB (embedded). stars: 15 · forks: 3 · watchers: 15 AI relevance: ai, agent, rag, mcp; High-value tags: agent, mcp, rag, eval; GitHub traction: 15 stars, 3 ...
Desktop AI workspace for leading models — separate sessions, built-in browser and tools, Studio documents and tables, scheduled tasks, plugins, and MCP. Powered by an open-source core. stars: 2 · watchers: 2 AI relevance: ai, llm, agent, mcp; High-value tags: agent, mcp, local, automation; GitHub traction: 2 stars, 0 forks
One web page plants a standing instruction in an AI agent's memory; sessions later, a clean request leaks email. Reproducible demo across Claude/GPT/Gemini/open-weights, plus agent-memgate: provenance-labelled memory + flow policy at the tool boundary (LangGraph & MCP adapters). AI relevance: ai, agent, mcp; High-value tags: agent, mcp; Freshness: 0.0h old
AI/ML & GenAI Developer | Building intelligent applications with Python, LLMs, RAG and Generative AI AI relevance: ai, llm, rag; High-value tags: rag; Freshness: 0.0h old
Open memory & reliability layer for AI agents + independent judging/verification (NACRE judge, caliber-bench meta-benchmark, preregistered criteria) — by Nautilus Platform stars: 1255 · forks: 36 · watchers: 1255 AI relevance: ai, llm, agent, rag; High-value tags: agent, mcp, rag, eval; GitHub traction: 1255 stars, 36 forks
AI agents for Unity — let a language model drive NPCs, factions and colonies, constrained to actions you register. Local-first, no cloud required. stars: 3 · watchers: 3 AI relevance: ai, llm, agent; High-value tags: agent, local; GitHub traction: 3 stars, 0 forks
Smithers is an agentic workflow framework for defining workflows in simple TypeScript configuration files and executing them quickly, durably, and reliably stars: 430 · forks: 51 · watchers: 430 AI relevance: ai, llm, agent; High-value tags: agent, workflow; GitHub traction: 430 stars, 51 forks
kb personal tools: MCPs (Google + Hostinger + ElevenLabs), Claude-driven email/planner workflows, daily calendar→Telegram digest AI relevance: ai, mcp; High-value tags: mcp, workflow; Freshness: 0.0h old
AI のための知識ベース。Wikipedia・OpenStreetMap・GeoNames の公式ダンプを SQLite (FTS5) に索引し、MCP と REST で AI に引かせる。AI が書き足せる notes と、Chiezo を道具として引くローカル LLM(agent モード・web 検索)も同梱。 AI relevance: ai, llm, agent, rag; High-value tags: agent, mcp, rag, local; Freshness: 0.0h old
Failure intelligence and shared debugging memory for AI coding agents AI relevance: ai, llm, agent, mcp; High-value tags: agent, coding agent, mcp; Freshness: 0.0h old
TypeScript SDK for durable AI agents: tool calling, approvals, sessions, MCP, and a CLI. Runs in your own Node process. stars: 7 · watchers: 7 AI relevance: ai, llm, agent, mcp; High-value tags: agent, mcp; GitHub traction: 7 stars, 0 forks
VENU.OS — AI/ML engineering portfolio featuring production RAG, multimodal AI, agentic systems, inference, and ML infrastructure. AI relevance: ai, agent, rag, inference; High-value tags: agent, rag, inference, multimodal; Freshness: 0.0h old
Retrieval-Augmented Legal Intelligence for Smart Legal Assistance DigiLaw is an AI-powered legal information and complaint assistance platform that uses Retrieval-Augmented Generation (RAG) to provide reliable, context-aware legal information from a curated knowledge base of legal documents. stars: 1 · watchers: 1 AI relevance: ai, rag, eval; High-value tags: rag, eval; GitHub traction: 1 stars, 0 forks
A multi-modal agent memory, context, inferencing, analytics engine and database. stars: 11 · forks: 2 · watchers: 11 AI relevance: agent; High-value tags: agent; GitHub traction: 11 stars, 2 forks
An open, programmable decision layer for models and compute. stars: 6048 · forks: 1016 · watchers: 6048 AI relevance: ai, llm, inference; High-value tags: inference; GitHub traction: 6048 stars, 1016 forks
Long-term memory for pi: recall what a session learned weeks later, by meaning, from the memsearch store Claude Code, Codex, OpenClaw and OpenCode share. stars: 3 · watchers: 3 AI relevance: agent, rag; High-value tags: agent, rag; GitHub traction: 3 stars, 0 forks
Production-style RAG assistant with document ingestion, vector search, reranking, citations, and evaluation. AI relevance: rag, eval; High-value tags: rag, eval, automation; Freshness: 0.0h old
Speech-first inference runtime in Rust — TTS / ASR / speech-to-speech / VC / speaker ID / VAD. An ONNX Runtime alternative that loads GGUF & safetensors directly: zero external dependencies, C ABI for Unity & Godot, CPU / Metal / CUDA / Vulkan backends. Apache-2.0, pre-release. stars: 13 · forks: 1 · watchers: 13 AI relevance: inference; High-value tags: inference; GitHub traction: 13 stars, 1 forks
SGLang-Omni is a high-performance serving framework for audio models (TTS, ASR) and unified multimodal models. stars: 1323 · forks: 557 · watchers: 1323 AI relevance: ai, inference, multimodal; High-value tags: inference, multimodal; GitHub traction: 1323 stars, 557 forks
Your AI intranet: network the computers you already own for inference and training. stars: 262 · forks: 13 · watchers: 262 AI relevance: ai, llm, inference; High-value tags: inference, local; GitHub traction: 262 stars, 13 forks
Tiny dependency-free C23 inference engine for small LLMs — ternary BitNet, CPU-first, runs on a Raspberry Pi stars: 3 · watchers: 3 AI relevance: ai, llm, inference; High-value tags: inference, local; GitHub traction: 3 stars, 0 forks
HN discussion: 1 points, 0 comments points: 1 AI relevance: ai, agent; High-value tags: agent; HN traction: 1 points, 0 comments
HN discussion: 1 points, 1 comments points: 1 · comments: 1 AI relevance: ai, rag; High-value tags: rag; HN traction: 1 points, 1 comments
HN discussion: 5 points, 2 comments points: 5 · comments: 2 AI relevance: rag; High-value tags: rag; HN traction: 5 points, 2 comments
HN discussion: 3 points, 1 comments points: 3 · comments: 1 AI relevance: rag; High-value tags: rag; HN traction: 3 points, 1 comments
HN discussion: 3 points, 1 comments points: 3 · comments: 1 AI relevance: ai, agent; High-value tags: agent; HN traction: 3 points, 1 comments
HN discussion: 3 points, 0 comments points: 3 AI relevance: ai, agent; High-value tags: agent; HN traction: 3 points, 0 comments
HN discussion: 10 points, 0 comments points: 10 AI relevance: inference; High-value tags: inference; HN traction: 10 points, 0 comments
HN discussion: 10 points, 0 comments points: 10 AI relevance: ai, mcp; High-value tags: mcp; HN traction: 10 points, 0 comments
HN discussion: 130 points, 62 comments points: 130 · comments: 62 AI relevance: llm; General trend signal; HN traction: 130 points, 62 comments
HN discussion: 43 points, 17 comments points: 43 · comments: 17 AI relevance: llm; High-value tags: local; HN traction: 43 points, 17 comments
Latent world models have shown a remarkable ability to predict future states and to plan in the real world. In practice, however, we lack a principled way to estimate how their capabilities scale with model size, data, and compute, an open problem that slows progress in the field. In this work we present RoboJEPA, a world model based on the Joint Embedding Predictive Architecture (JEPA) and trained on a large-scale d...
Language-model agents are increasingly asked to carry out open-ended scientific research, yet their results are usually graded against a known answer, a rubric, or a language-model reviewer, none of which can tell whether a new scientific model is valid. The AI Science Exam for El Nino-Southern Oscillation (SciExam for ENSO) is a benchmark in which agents build low-order stochastic models of ENSO, the dominant mode o...
Large language models are increasingly applied to tasks grounded in long, heterogeneous information sources. Conventional Retrieval-Augmented Generation (RAG) relies on fixed similarity-based retrieval, while agentic variants adapt queries and tool use but remain largely retrieval-centric. However, in many tasks, the evidence required for a solution is not explicitly present in any single source item. Instead, it mus...