Report Date: 2026-02-07
Analysis Window: 30 days
Languages Tracked: python, typescript, rust, go, javascript
Top N: 50
Overview
近期 GitHub AI 开源领域趋势明确，AI Agent（智能体）成为绝对焦点，占据了榜单的大部分席位。技术栈上，TypeScript 保持领先地位，而 Python 紧随其后。整体趋势显示，社区正集中力量构建和优化各类编码 Agent、Agent 记忆系统及协同工作流。
Top 50 Repositories
| Rank | Repository | Category | Language | Heat_7 | Heat_30 | Score |
|---|---|---|---|---|---|---|
| 1 | openclaw | typescript | 58,305 | 168,498 | 3,060 | |
| 2 | opencode | agent | typescript | 0 | 48,566 | 161 |
| 3 | skills | agent | python | 0 | 30,541 | 449 |
| 4 | oh-my-opencode | agent | typescript | 0 | 19,658 | 65 |
| 5 | claude-mem | agent | typescript | 8,838 | 13,497 | 423 |
| 6 | PageIndex | rag | python | 3,051 | 9,233 | 161 |
| 7 | vibe-kanban | agent | typescript | 0 | 7,448 | 24 |
| 8 | beads | agent | go | 0 | 6,747 | 22 |
| 9 | UI-TARS-desktop | agent | typescript | 2,211 | 6,675 | 287 |
| 10 | pi-mono | agent | typescript | 3,892 | 6,290 | 187 |
| 11 | dexter | agent | typescript | 0 | 4,773 | 15 |
| 12 | memU | agent | python | 1,737 | 4,704 | 90 |
| 13 | goose | agent | rust | 0 | 4,610 | 15 |
| 14 | codex | agent | rust | 1,182 | 4,278 | 156 |
| 15 | shannon | agent | typescript | 3,560 | 4,246 | 2,046 |
| 16 | context7 | llm | typescript | 0 | 4,084 | 13 |
| 17 | openskills | agent | typescript | 0 | 3,878 | 12 |
| 18 | ChatLab | agent | typescript | 0 | 3,045 | 10 |
| 19 | crush | agent | go | 0 | 2,881 | 9 |
| 20 | UltraRAG | rag | python | 0 | 2,726 | 9 |
| 21 | BitNet | llm | python | 1,159 | 2,725 | 58 |
| 22 | kimi-cli | agent | python | 932 | 2,368 | 47 |
| 23 | supermemory | typescript | 0 | 2,296 | 7 | |
| 24 | memvid | rag | rust | 0 | 2,089 | 6 |
| 25 | ralph-orchestrator | agent | rust | 0 | 1,410 | 4 |
| 26 | AIClient-2-API | llm | javascript | 517 | 1,267 | 137 |
| 27 | OpenAgentsControl | agent | typescript | 0 | 1,188 | 3 |
| 28 | claude-skills | llm | python | 0 | 1,122 | 3 |
| 29 | ext-apps | typescript | 0 | 1,070 | 3 | |
| 30 | worktrunk | agent | rust | 0 | 1,046 | 3 |
| 31 | meeting-minutes | rust | 0 | 641 | 2 | |
| 32 | bifrost | llm | go | 0 | 547 | 1 |
| 33 | agent-lightning | agent | python | 2,270 | 0 | 97 |
| 34 | cognee | agent | python | 518 | 0 | 61 |
| 35 | easy-dataset | llm | javascript | 164 | 0 | 58 |
| 36 | Auto-Claude | agent | typescript | 0 | 0 | 54 |
| 37 | Maestro | agent | typescript | 1,123 | 0 | 48 |
| 38 | upscayl | vision | typescript | 0 | 0 | 46 |
| 39 | mem0 | agent | python | 0 | 0 | 45 |
| 40 | chrome-devtools-mcp | agent | typescript | 0 | 0 | 43 |
| 41 | RD-Agent | python | 335 | 0 | 35 | |
| 42 | plano | agent | rust | 0 | 0 | 33 |
| 43 | 1Panel | llm | go | 0 | 0 | 25 |
| 44 | refly | agent | typescript | 0 | 0 | 23 |
| 45 | review-prompts | python | 486 | 0 | 20 | |
| 46 | vibetunnel | agent | typescript | 477 | 0 | 20 |
| 47 | AgentGPT | agent | typescript | 0 | 0 | 9 |
| 48 | UI-TARS | agent | python | 196 | 0 | 8 |
| 49 | axonhub | llm | go | 201 | 0 | 8 |
| 50 | kagent | agent | go | 0 | 0 | 7 |
Category Breakdown
rag (3 projects)
检索增强生成（RAG）领域虽然项目不多（3个），但呈现出技术演进的趋势。项目如 PageIndex 探索“无向量、基于推理”的 RAG 方案，显示出对传统 RAG 架构的创新思考。
Average Heat_7: 1,017
Average Score: 58
- PageIndex - 📑 PageIndex: Document Index for Vectorless, Reasoning-based RAG
- UltraRAG - UltraRAG v3: A Low-Code MCP Framework for Building Complex and Innovative RAG Pipelines
- memvid - Memory layer for AI Agents. Replace complex RAG pipelines with a serverless, single-file memory layer. Give your agents instant retrieval and long-term memory.
llm (8 projects)
大语言模型领域依然重要，有8个项目上榜。关注点主要集中在模型服务网关（如 bifrost）、高效推理框架（如微软的 BitNet）以及数据集构建工具（easy-dataset）。
Average Heat_7: 255
Average Score: 37
- context7 - Context7 MCP Server -- Up-to-date code documentation for LLMs and AI code editors
- BitNet - Official inference framework for 1-bit LLMs
- AIClient-2-API - Simulates Gemini CLI, Antigravity, Qwen Code, and Kiro client requests, compatible with the OpenAI API. It supports thousands of Gemini model requests per day and offers free use of the built-in Cl...
- claude-skills - A Collection of Skills for Claude Code and Claude AI for real-world Usage. Including Claude Code Subagents, Claude Code Commnads
- bifrost - Fastest LLM gateway (50x faster than LiteLLM) with adaptive load balancer, cluster mode, guardrails, 1000+ models support & <100 µs overhead at 5k RPS.
- easy-dataset - A powerful tool for creating datasets for LLM fine-tuning 、RAG and Eval
- 1Panel - 🔥 1Panel offers an intuitive web interface for Linux server, making it easy to manage OpenClaw agents, LLMs, websites, databases, containers, files, and scheduled tasks.
- axonhub - ⚡️ Open-source AI Gateway — Use any SDK to call 100+ LLMs. Built-in failover, load balancing, cost control & end-to-end tracing.
vision (1 projects)
视觉领域在此次榜单中仅有一个项目（upscayl），表明与 Agent 相关的开发热度暂时超过了传统的 AI 视觉应用。
Average Heat_7: 0
Average Score: 46
- upscayl - 🆙 Upscayl - #1 Free and Open Source AI Image Upscaler for Linux, MacOS and Windows.
(6 projects)
该类别虽然项目少，但平均热度和评分极高，这主要是由排名第一的 openclaw/openclaw 项目驱动。这些项目通常是目标宏大的通用 AI 助手或底层框架。
Average Heat_7: 9,854
Average Score: 521
- openclaw - Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞
- supermemory - Memory engine and app that is extremely fast, scalable. The Memory API for the AI era.
- ext-apps - Official repo for spec & SDK of MCP Apps protocol - standard for UIs embedded AI chatbots, served by MCP servers
- meeting-minutes - Privacy first, AI meeting assistant with 4x faster Parakeet/Whisper live transcription, speaker diarization, and Ollama summarization built on Rust. 100% local processing. no cloud required. Meetil...
- RD-Agent - Research and development (R&D) is crucial for the enhancement of industrial productivity, especially in the AI era, where the core aspects of R&D are mainly focused on data and models. We are commi...
- review-prompts - AI review prompts
agent (32 projects)
智能体是本次趋势的核心，50个项目中独占32席。开发者对自主编码、金融研究、安全渗透等领域的 Agent 表现出极大热情，相关基础设施和工具链的开发异常活跃。
Average Heat_7: 841
Average Score: 139
- opencode - The open source coding agent.
- skills - Public repository for Agent Skills
- oh-my-opencode - The Best Agent Harness. Meet Sisyphus: The Batteries-Included Agent that codes like you.
- claude-mem - A Claude Code plugin that automatically captures everything Claude does during your coding sessions, compresses it with AI (using Claude's agent-sdk), and injects relevant context back into future ...
- vibe-kanban - Get 10X more out of Claude Code, Codex or any coding agent
- beads - Beads - A memory upgrade for your coding agent
- UI-TARS-desktop - The Open-Source Multimodal AI Agent Stack: Connecting Cutting-Edge AI Models and Agent Infra
- pi-mono - AI agent toolkit: coding agent CLI, unified LLM API, TUI & web UI libraries, Slack bot, vLLM pods
- dexter - An autonomous agent for deep financial research
- memU - Memory for 24/7 proactive agents like openclaw (moltbot, clawdbot).
- goose - an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM
- codex - Lightweight coding agent that runs in your terminal
- shannon - Fully autonomous AI hacker to find actual exploits in your web apps. Shannon has achieved a 96.15% success rate on the hint-free, source-aware XBOW Benchmark.
- openskills - Universal skills loader for AI coding agents - npm i -g openskills
- ChatLab - A Local-first chat analysis tool: Relive your social memories powered by SQL and AI Agents. 本地化的聊天记录分析工具，通过 SQL 和 AI Agent 回顾你的社交记忆。
- crush - Glamourous agentic coding for all 💘
- kimi-cli - Kimi Code CLI is your next CLI agent.
- ralph-orchestrator - An improved implementation of the Ralph Wiggum technique for autonomous AI agent orchestration
- OpenAgentsControl - AI agent framework for plan-first development workflows with approval-based execution. Multi-language support (TypeScript, Python, Go, Rust) with automatic testing, code review, and validation buil...
- worktrunk - Worktrunk is a CLI for Git worktree management, designed for parallel AI agent workflows
- agent-lightning - The absolute trainer to light up AI agents.
- cognee - Memory for AI Agents in 6 lines of code
- Auto-Claude - Autonomous multi-session AI coding
- Maestro - Agent Orchestration Command Center
- mem0 - Universal memory layer for AI Agents
- chrome-devtools-mcp - Chrome DevTools for coding agents
- plano - Delivery infrastructure for agentic apps - Plano is an AI-native proxy and data plane that offloads plumbing work, so you stay focused on your agent's core logic (via any AI framework).
- refly - The first open-source agent skills builder. Define skills by vibe workflow, run on Claude Code, Cursor, Codex & more. Build Clawdbot 🦞· APIs for Lovable · Bots for Slack & Lark/Feishu · Skill...
- vibetunnel - Turn any browser into your terminal & command your agents on the go.
- AgentGPT - 🤖 Assemble, configure, and deploy autonomous AI Agents in your browser.
- UI-TARS - Pioneering Automated GUI Interaction with Native Agents
- kagent - Cloud Native Agentic AI | Discord: https://bit.ly/kagentdiscord
Dark Horse Projects
“黑马”项目表现抢眼，特别是 openclaw/openclaw（得分 3060）和 KeygraphHQ/shannon（得分 2046），它们的得分远超其他项目。这表明其新颖的定位（个人 AI 助手、自主AI黑客）迅速获得了社区的高度关注和认可。此外，anthropics/skills 和 thedotmack/claude-mem 等项目也凭借其在 Agent 能力扩展和记忆方面的创新获得了强劲的初始动力。
| Repository | Score | Heat_30 | Heat_7 | Category |
|---|---|---|---|---|
| openclaw | 3,060 | 168,498 | 58,305 | |
| opencode | 161 | 48,566 | 0 | agent |
| skills | 449 | 30,541 | 0 | agent |
| claude-mem | 423 | 13,497 | 8,838 | agent |
| PageIndex | 161 | 9,233 | 3,051 | rag |
| UI-TARS-desktop | 287 | 6,675 | 2,211 | agent |
| pi-mono | 187 | 6,290 | 3,892 | agent |
| codex | 156 | 4,278 | 1,182 | agent |
| shannon | 2,046 | 4,246 | 3,560 | agent |
| AIClient-2-API | 137 | 1,267 | 517 | llm |
Consecutive Appearances
所有50个上榜项目均为连续上榜，这是一个非常强烈的信号，表明当前 AI Agent 的热潮并非昙花一现，而是正在形成稳固的技术趋势。社区对编码 Agent、Agent 记忆和任务编排等方向的兴趣正在持续巩固，相关项目获得了稳定的关注度。
| Repository | Weeks in Top | Category | Current Heat_7 |
|---|---|---|---|
| openclaw | 2 | 58,305 | |
| opencode | 2 | agent | 0 |
| skills | 2 | agent | 0 |
| oh-my-opencode | 2 | agent | 0 |
| claude-mem | 2 | agent | 8,838 |
| PageIndex | 2 | rag | 3,051 |
| vibe-kanban | 2 | agent | 0 |
| beads | 2 | agent | 0 |
| UI-TARS-desktop | 2 | agent | 2,211 |
| pi-mono | 2 | agent | 3,892 |
| dexter | 2 | agent | 0 |
| memU | 2 | agent | 1,737 |
| goose | 2 | agent | 0 |
| codex | 2 | agent | 1,182 |
| shannon | 2 | agent | 3,560 |
| context7 | 2 | llm | 0 |
| openskills | 2 | agent | 0 |
| ChatLab | 2 | agent | 0 |
| crush | 2 | agent | 0 |
| UltraRAG | 2 | rag | 0 |
| BitNet | 2 | llm | 1,159 |
| kimi-cli | 2 | agent | 932 |
| supermemory | 2 | 0 | |
| memvid | 2 | rag | 0 |
| ralph-orchestrator | 2 | agent | 0 |
| AIClient-2-API | 2 | llm | 517 |
| OpenAgentsControl | 2 | agent | 0 |
| claude-skills | 2 | llm | 0 |
| ext-apps | 2 | 0 | |
| worktrunk | 2 | agent | 0 |
| meeting-minutes | 2 | 0 | |
| bifrost | 2 | llm | 0 |
| agent-lightning | 2 | agent | 2,270 |
| cognee | 2 | agent | 518 |
| easy-dataset | 2 | llm | 164 |
| Auto-Claude | 2 | agent | 0 |
| Maestro | 2 | agent | 1,123 |
| upscayl | 2 | vision | 0 |
| mem0 | 2 | agent | 0 |
| chrome-devtools-mcp | 2 | agent | 0 |
| RD-Agent | 2 | 335 | |
| plano | 2 | agent | 0 |
| 1Panel | 2 | llm | 0 |
| refly | 2 | agent | 0 |
| review-prompts | 2 | 486 | |
| vibetunnel | 2 | agent | 477 |
| AgentGPT | 2 | agent | 0 |
| UI-TARS | 2 | agent | 196 |
| axonhub | 2 | llm | 201 |
| kagent | 2 | agent | 0 |
Highlighted Repositories
openclaw/openclaw
作为榜单的绝对领先者，其热度和得分远超其他项目。项目定位为“你自己的个人 AI 助手，支持任何操作系统和平台”，显示出其构建通用型智能体的宏大目标。它也是未分类项目中各项指标的绝对贡献者。
anomalyco/opencode
该项目与 anthropics/skills 共同定义了本次趋势的核心主题——编码 Agent。opencode 是一个开源编码 Agent 的实现，虽然近期7日热度为0，但凭借在30天窗口期前段积累的巨大优势稳居高位，反映了市场对开 源编码工具的强烈需求。
KeygraphHQ/shannon
该项目因其独特的应用场景而备受关注——“全自主的 AI 黑客”，用于发现 Web 应用中的真实漏洞。其高达 2046 的分数说明，AI Agent 在网络安全这一垂直领域的高价值应用已引起了开发者的极大兴趣。
VectifyAI/PageIndex
作为 RAG 领域的代表，该项目提出了“无向量、基于推理的 RAG”概念。这预示着 RAG 技术可能正从依赖向量相似度搜索，向结合更多逻辑推理能力的方向演进，是一个值得关注的技术信号。
bytedance/UI-TARS-desktop
来自字节跳动的该项目，专注于“开源多模态 AI Agent 栈”，旨在实现自动化的 GUI 交互。这标志着 Agent 的能力正在从纯文本和代码环境，扩展到更复杂的图形用户界面操作，是 Agent 应用走向更广泛场景的关键一步。
Methodology
Data Sources:
- GitHub Trending pages
- GitHub API stargazer timestamps
Metrics:
- Heat_7: Stars gained in last 7 days
- Heat_30: Stars gained in last 30 days
- Score: Weighted scoring combining short-term heat and sustained growth
- Formula: 0.6 × stars_1d + 0.3 × (stars_7d / 7) + 0.1 × (stars_30d / 30)
- Emphasizes recent activity (60%) while considering sustained trends (40%)
Filtering:
- Include keywords: %!s(int=30)
- Exclude keywords: %!s(int=30)
- Categories: %!s(int=30)
Ranking:
- Sort by Heat_30 (descending)
- Tie-break by Score (descending)
- Select top %!d(string=ai, llm, agent, gpt, rag, langchain, machine-learning, deep-learning, neural, transformer, embedding, vector)
Limitations:
- Trending data limited to GitHub's trending algorithm
- Star history may be incomplete for repos with >40k stars (API pagination limits)
- LLM-generated commentary is interpretive, not prescriptive
- Weekly snapshots may miss short-lived trends%!(EXTRA string=tutorial, awesome, learning, course, book, string=agent, llm, rag, vision, infra, int=50)
Generated at: 2026-02-07T15:47:04Z