构建、评估并优化 AI 系统。涵盖 evals、RAG、agents、微调、合成数据生成、数据集管理、MCP 等内容。Build, Evaluate, and Optimize AI Systems. Includes evals, RAG, agents, fine-tuning, synthetic data generation, dataset management, MCP, and more.
趋势榜
实时增长榜 · 周综述每周日由 spark 生成
按日期回看
2026-09-02 新收录仓库 44 个
AG2(原 AutoGen):开源 AgentOS。欢迎加入:https://discord.gg/sNGSwQME3xAG2 (formerly AutoGen): The Open-Source AgentOS.Join us at: https://discord.gg/sNGSwQME3x
在未集成任何 DLSS 的 D3D11/D3D12/Vulkan 游戏中启用 DLSS 5 神经渲染,通过私有 D3D12 设备将合成 DLAA 契约(ReShade 深度 + 运动矢量)送入 DLSS 5 插件。DLSS 5 neural rendering in D3D11/D12/Vulkan games that ship without any DLSS — feeds a synthetic DLAA contract (ReShade depth + motion vectors) to the DLSS 5 add-on via a private D3D12 device.
面向自进化 agent 的持续学习基础设施。Continual learning infra for self-improving agents
🛡️ AI agents 的审批与策略层。在高风险动作执行前进行拦截、阻断或远程审批。🛡️ The approval and policy layer for AI agents. Intercept risky actions before they run, block them, or approve them remotely.
多 Agent 应用,支持并行会话、工具、记忆、工作流,覆盖桌面、终端与移动端。Multi-agent app for parallel sessions, tools, memory, workflows, desktop, terminal, and mobile.
在 Mac 上以远低于 104 GB 的内存运行 Qwen3.8-Flash-Next(125B MoE,4-bit 下约 104 GB),通过 SSD 流式加载专家。MLX + Swift,兼容 Ollama 的 API。Run Qwen3.8-Flash-Next (125B MoE, 104 GB at 4-bit) on Macs with a fraction of that RAM by streaming experts from SSD. MLX + Swift, Ollama-compatible API.
将 YouTube、Bilibili 或本地课程/会议录像转换为带幻灯片配图的 Markdown 与 HTML 讲义。Turn YouTube, Bilibili, or local course/meeting recordings into slide-illustrated Markdown and HTML lecture notes.
面向 OpenAI Codex CLI 的自治系统与安全工具集。确定性二进制分析、解耦流水线与零摩擦执行。Autonomous Systems & Security Toolkit for OpenAI Codex CLI. Deterministic binary analysis, decoupled pipeline, and zero-friction execution.
一键搭建泄漏版 DLSS 5 神经渲染构建版,适用于任何 DX11/DX12 游戏,无论是否已集成 DLSS。使用 ReShade + RenoDX 插件(或 OptiScaler 引擎);针对无 DLSS 游戏提供 DLSS5-Feeder + LumeniteFX;针对 DX11 提供 dlss5-bridge。Rust 实现,单个 exe。One-click setup of the leaked DLSS 5 neural-rendering build for any DX11/DX12 game on RTX 20–50, with or without DLSS. ReShade + RenoDX add-on (or OptiScaler engine); DLSS5-Feeder + LumeniteFX for games without DLSS; dlss5-bridge for DX11. Rust, single exe.
开源的 Windows 按应用 AI 辅助分流方案,提供 OpenAI/Ollama 草稿生成与 sing-box TUN 数据面。Open-source AI-assisted per-application routing for Windows with OpenAI/Ollama drafts and a sing-box TUN data plane.
面向你的 GitHub 个人主页的、隐私优先的 AI 编程活动。Privacy-first AI coding activity for your GitHub profile.
⚡ 原生 LLM、自进化的量化交易终端与执行引擎(彭博风格深色终端 + 六资产 AI 大脑)。⚡ LLM-Native Self-Evolving Quantitative Trading Terminal & Execution Engine (Bloomberg-Style Dark Terminal + 6-Asset AI Brain)
本地 AI 注册中心——硬件、模型、recipes、模型实例与价格。Local AI registry — hardware, models, recipes, model instances, and prices
SmartBus 是一个基于 Web 的公交追踪与管理系统,旨在简化公共交通管理。用户可查看公交线路、站点、时刻表并实时追踪车辆位置。系统还包含管理模块,以便高效管理车辆、线路、站点、时刻表及追踪信息。SmartBus is a web-based Bus Tracking and Management System developed to simplify public transportation management. The application allows users to view bus routes, bus stops, bus timings, and track bus locations. The system also includes an admin module to manage buses, routes, stops, timings, and tracking information efficiently.
300+ 全新可用 captive portal 列表,专为 marauder 固件制作,文件均为 .html 格式。A list of over 300+ fresh, working captive portals specifically made for the marauder firmware. Files are all in .html format.
面向视觉创作者与编程 agent 的本地音频响应节奏图与可复用时序包。Local audio-reactive rhythm maps and reusable timing packages for visual creators and coding agents.
面向 AI agent 的本地优先、证据可控的学术写作工作流,具备受限修订、净室评审与发布治理能力。Local-first, evidence-controlled academic writing workflows for AI agents, with bounded revision, clean-room review, and release governance.
按能力路由、证据优先的 Agent Skill,覆盖构思、文献、统计、图表、写作、同行评审、rebuttal、论文以及论文转 PPT。Capability-routed, evidence-first Agent Skill for ideation, literature, statistics, figures, writing, peer review, rebuttals, theses, and paper-to-PPT.
Noetrium:面向 AI agent 的可复现研究基础设施 —— 涵盖实验、模型、环境、证据、恢复、评估与验证。Noetrium: Reproducible Research Infrastructure for AI Agents - experiments, models, environments, evidence, recovery, evaluation, and verification.
AnchorBell:以 Rust 为先、仅做 Maker 的 Binance 股票永续交易引擎 —— 可复现研究、历史回测、市场数据回放、受控的 Testnet 执行、风险控制、订单生命周期管理、恢复机制与可观测性。AnchorBell: Rust-First Maker-Only Binance Equity Perpetual Trading Engine - reproducible research, historical backtesting, market-data replay, controlled Testnet execution, risk controls, order lifecycle management, recovery, and observability.
Noetrium:面向 AI Agent 的可复现研究基础设施——实验、模型、环境、证据、恢复、评估与验证。Noetrium: Reproducible Research Infrastructure for AI Agents - experiments, models, environments, evidence, recovery, evaluation, and verification.
AnchorBell:Rust 优先、仅做市商模式的 Binance 股票永续交易引擎——可复现研究、历史回测、行情数据重放、受控 Testnet 执行、风险控制、订单生命周期管理、恢复与可观测性。AnchorBell: Rust-First Maker-Only Binance Equity Perpetual Trading Engine - reproducible research, historical backtesting, market-data replay, controlled Testnet execution, risk controls, order lifecycle management, recovery, and observability.
面向学术预印本的投稿前审查:对每条引用在 Crossref、Semantic Scholar、arXiv 与 OpenAlex 上进行核验,并对 .tex 源码执行确定性的 desk-reject 守卫。Pre-submission review for academic preprints: verifies every citation against Crossref, Semantic Scholar, arXiv, and OpenAlex, and runs deterministic desk-reject guards on the .tex source.
按分支整理的北大课程存档,含精选报告与学术写作作品集。PKU course archives organized by branch, with a curated portfolio of selected reports and academic writing.
一个面向应用语言学学术写作的、注重隐私安全且严格依据证据的 Codex skill。A privacy-safe, evidence-bound Codex skill for applied-linguistics academic writing.
算数量子力学的可复现研究实验记录。Reproducible research lab book for arithmetic quantum mechanics
开箱即用的 GUI 工具,面向人在环上监督的 AI 辅助系统综述写作。可在一站式应用中协同、管理、审阅与核查综述资料来源与信息。Ready-to-install GUI for human-in-the-loop AI assisted systematic review writing. Allows coordinating, managing, reviewing, and verifying review sourcing and information all from one easy to use app.
检索 EndNote XML 元数据,保守地匹配本地 PDF,并以可追溯的证据回答研究问题。Search EndNote XML metadata, conservatively match local PDFs, and answer research questions with traceable evidence.
基于 Neo4j 知识图谱回答生物医学研究问题的多 Agent 系统——采用 LangGraph 编排、图+向量混合 RAG、只读 Cypher 护栏、Docker 隔离的代码执行与人在环上审核。Multi-agent system answering biomedical research questions over a Neo4j knowledge graph — LangGraph orchestration, hybrid graph+vector RAG, read-only Cypher guardrails, Docker-isolated code execution and human-in-the-loop review
基于 LangGraph 的 AI Agent,分析上市公司季度财报,利用 SEC 文件与结构化财务数据,帮助用户回答限定范围内的投研问题。A LangGraph-based AI agent that analyzes public company quarterly earnings reports and helps users answer a limited set of investment-research questions using SEC filings and structured financial data.
经济研究、计量与统计分析及数据项目作品集。Portfolio of economic research, econometric and statistical analysis, and data projects.
一个面向优化与控制领域论文的、基于语料库驱动的学术写作润色 skill。A corpus-driven academic writing polishing skill for optimization and control papers.
开放研究项目:对乳腺癌、肺癌与黑色素瘤领域近期肿瘤学临床试验进行 meta 分析。Open research project: Meta-analysis of recent oncology clinical trials across breast cancer, lung cancer, and melanoma
支持神经保留 RARP 术中热规避与能量使用系统综述及 meta 分析的衍生数据与 R 代码。Derived data and R code supporting a systematic review and meta-analysis of thermal avoidance versus energy use during nerve-sparing RARP.
多 Agent 研究编排框架,覆盖文献检索、论文分析、证据抽取、跨论文对比、知识综合与经过验证的研究空白发现。A multi-agent research orchestration framework for literature retrieval, paper analysis, evidence extraction, cross-paper comparison, knowledge synthesis, and validated research gap discovery.
2026-09-02 增长榜定格 当日周增 Top 20
| 名次 | 名称 | 方向 | 周增 | Stars |
|---|---|---|---|---|
| 1 — | odysseus-dev/odysseus | Agent 智能体 | +152,621 | 84,138 |
| 2 — | anywhere-labs/deepseek-harness-desktop | 评测基准 | +8,281 | 13,495 |
| 3 — | stablyai/orca | Agent 智能体 | +5,299 | 47,228 |
| 4 — | s1dashu/ip-as-logo-skill | Agent 智能体 | +4,646 | 3,727 |
| 5 — | MiniMax-AI/MiniMax-H3 | LLM 基础设施 | +4,445 | 4,611 |
| 6 — | diegosouzapw/OmniRoute | Agent 智能体 | +4,343 | 53,672 |
| 7 — | holaboss-ai/holaOS | Agent 智能体 | +3,798 | 9,436 |
| 8 — | safishamsi/graphify | 数据与向量库 | +3,752 | 76,856 |
| 9 — | NousResearch/hermes-agent | Agent 智能体 | +3,386 | 230,735 |
| 10 — | zhu1090093659/dsh-web-ui | 评测基准 | +3,318 | 4,541 |
| 11 — | firecrawl/firecrawl | RAG 检索增强 | +2,975 | 165,271 |
| 12 — | firecrawl/anydoc | 工程化 | +2,870 | 13,720 |
| 13 — | koala73/worldmonitor | Agent 智能体 | +2,239 | 82,478 |
| 14 — | hoainho/img2threejs | 多模态 | +2,051 | 4,176 |
| 15 — | bojieli/ai-agent-book | Agent 智能体 | +1,978 | 43,727 |
| 16 — | DietrichGebert/ponytail | Agent 智能体 | +1,869 | 100,293 |
| 17 新上榜 | Nanako0129/sepia | Agent 智能体 | +1,769 | 1,345 |
| 18 ↓1 | anthropics/skills | Agent 智能体 | +1,722 | 167,649 |
| 19 ↓1 | infiniflow/ragflow | RAG 检索增强 | +1,661 | 88,637 |
| 20 ↓1 | addyosmani/agent-skills | Agent 智能体 | +1,631 | 85,866 |