agentic-ai-projects.csv

repo_id,repo_name,description,stars,contributors,forks,open_issues,license,archived,pushed_at,openrank_2607,openrank_2608,openrank_trend_2508_2607,openrank_trend_2509_2608,participants_2607,language,created_at,topics,landscape_action,landscape_layer,landscape_section,selection_reason,selection_caveat,github_status,trend_signal,trend_signal_reason
1103012935,openclaw/openclaw,The AI that really does things. Any OS. Any Platform. The lobster way. 🦞 ,390374,377,82130,8517,NOASSERTION,false,2026-09-24,462.71,303.24,"[null,null,null,0.92,10.91,497.55,1693.69,1871.65,1499.86,880.67,592.19,462.71]","[null, null, 0.92, 10.91, 497.55, 1693.69, 1871.65, 1499.86, 880.67, 592.19, 462.71, 303.24]",281,TypeScript,2025-11-24,"ai,assistant,crustacean,molty,openclaw,own-your-data,personal",keep,Agent Infra,Personal AI assistants,"当前图中的 Personal AI assistants 代表项目；仓库仍在维护，384,408 stars，最近可用 OpenRank 631.92。",产品边界普遍较宽，只在这一分区出现，避免与 chatbot 和 framework 重复。,ok,,
626805178,langgenius/dify,"Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.",157081,454,24760,1101,NOASSERTION,false,2026-09-24,53.65,49.17,"[387.79,385.95,326.28,286.14,287.31,266.36,217.58,197.49,145.19,104.87,77.64,53.65]","[385.95, 326.28, 286.14, 287.31, 266.36, 217.58, 197.49, 145.19, 104.87, 77.64, 53.65, 49.17]",39,TypeScript,2023-04-12,"agent,agentic-ai,agentic-framework,agentic-workflow,ai,automation,claude,deepseek,genai,gpt,llm,low-code,mcp,nextjs,no-code,openai,python,skills,workflow",keep,Agent Infra,Workflow & agent builders,"当前图中的 Workflow & agent builders 代表项目；仓库仍在维护，150,556 stars，最近可用 OpenRank 85.63。",与通用低代码平台有交叉，主图只保留明确支持 agent workflow 的项目。,ok,,
975734319,anomalyco/opencode,The open source coding agent.,209824,456,27713,6210,MIT,false,2026-09-24,157.72,118.02,"[283.25,257.3,263.91,315.9,377.89,729.96,681.2,553.3,468.82,386.73,248.46,157.72]","[257.3, 263.91, 315.9, 377.89, 729.96, 681.2, 553.3, 468.82, 386.73, 248.46, 157.72, 118.02]",97,TypeScript,2025-04-30,,keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，190,463 stars，最近可用 OpenRank 271.45。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
965415649,openai/codex,Lightweight coding agent that runs in your terminal,126313,475,19695,18598,Apache-2.0,false,2026-09-24,175.4,80.64,"[251.55,423.95,391.38,407.9,373.89,365.47,468.81,479.16,469.03,457.04,334.43,175.4]","[423.95, 391.38, 407.9, 373.89, 365.47, 468.81, 479.16, 469.03, 457.04, 334.43, 175.4, 80.64]",122,Rust,2025-04-13,,keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，102,090 stars，最近可用 OpenRank 373.91。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
968197216,google-gemini/gemini-cli,An open-source AI agent that brings the power of Gemini directly into your terminal.,107145,443,14629,825,Apache-2.0,false,2026-09-24,35.41,27.16,"[542.58,498.43,444.22,421.48,370.75,371.57,385.74,428.49,344.39,232.46,84.58,35.41]","[498.43, 444.22, 421.48, 370.75, 371.57, 385.74, 428.49, 344.39, 232.46, 84.58, 35.41, 27.16]",11,TypeScript,2025-04-17,"ai,ai-agents,cli,gemini,gemini-api,mcp-client,mcp-server",keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，106,219 stars，最近可用 OpenRank 93.64。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
193215554,n8n-io/n8n,"Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.",205856,428,60872,1154,NOASSERTION,false,2026-09-24,82.05,84.38,"[328.31,329.51,306.99,299.67,295.84,285.89,254.98,229.22,209.75,190.55,131.52,82.05]","[329.51, 306.99, 299.67, 295.84, 285.89, 254.98, 229.22, 209.75, 190.55, 131.52, 82.05, 84.38]",51,TypeScript,2019-06-22,"ai,apis,automation,cli,data-flow,development,integration-framework,integrations,ipaas,low-code,low-code-platform,mcp,mcp-client,mcp-server,n8n,no-code,self-hosted,typescript,workflow,workflow-automation",keep,Agent Infra,Workflow & agent builders,"当前图中的 Workflow & agent builders 代表项目；仓库仍在维护，198,383 stars，最近可用 OpenRank 156.06。",与通用低代码平台有交叉，主图只保留明确支持 agent workflow 的项目。,ok,,
671269505,BerriAI/litellm,"The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM]",59551,374,11741,5240,NOASSERTION,false,2026-09-24,92.22,71.18,"[321.12,312.79,288.67,257.21,255.62,259.11,277.86,244.03,183.69,141.15,111.17,92.22]","[312.79, 288.67, 257.21, 255.62, 259.11, 277.86, 244.03, 183.69, 141.15, 111.17, 92.22, 71.18]",121,Python,2023-07-27,"ai-gateway,anthropic,azure-openai,bedrock,gateway,langchain,litellm,llm,llm-gateway,llmops,mcp-gateway,openai,openai-proxy,rust,rust-ai,vertex-ai",keep,Model Infra,Model API gateways,"当前图中的 Model API gateways 代表项目；仓库仍在维护，54,916 stars，最近可用 OpenRank 113.81。",需要区分 model API、MCP gateway 和 agentic proxy，避免把所有代理工具放在一起。,ok,,
1061953414,anthropics/skills,Public repository for Agent Skills,177924,15,21078,1284,,false,2026-09-22,,7.05,"[null,null,6.01,7.29,8.21,13.65,15.76,14.27,16.12,9.92,5.13,null]","[null, 6.01, 7.29, 8.21, 13.65, 15.76, 14.27, 16.12, 9.92, 5.13, null, 7.05]",6,Python,2025-09-22,agent-skills,keep,Agent Infra,Protocols & interoperability,"当前图中的 Protocols & interoperability 代表项目；仓库仍在维护，164,751 stars，最近可用 OpenRank 4.95。",协议成熟度差异较大，preview 或规范状态需要在图上明确。,ok,,
890668799,modelcontextprotocol/servers,Model Context Protocol Servers,90575,427,11689,552,NOASSERTION,false,2026-09-22,8.13,6.11,"[66.04,48.42,34.35,30.6,19.34,16.12,14.73,17.78,19.79,13.22,8.62,8.13]","[48.42, 34.35, 30.6, 19.34, 16.12, 14.73, 17.78, 19.79, 13.22, 8.62, 8.13, 6.11]",15,TypeScript,2024-11-19,,keep,Agent Infra,Protocols & interoperability,"当前图中的 Protocols & interoperability 代表项目；仓库仍在维护，88,989 stars，最近可用 OpenRank 7.99。",协议成熟度差异较大，preview 或规范状态需要在图上明确。,ok,,
881458615,browser-use/browser-use,Agents that use the browser.,116169,361,12791,491,MIT,false,2026-09-24,5.4,9.63,"[98.41,94.44,70.85,61.76,53.3,42.84,37.69,31.22,20.23,15.16,12.74,5.4]","[94.44, 70.85, 61.76, 53.3, 42.84, 37.69, 31.22, 20.23, 15.16, 12.74, 5.4, 9.63]",15,Python,2024-10-31,"ai-agents,ai-tools,browser-automation,browser-use,llm,playwright,python",keep,Agent Infra,"Tools, web & computer use","当前图中的 Tool & browser use 代表项目；仓库仍在维护，107,103 stars，最近可用 OpenRank 9.58。",browser/computer-use 项目迭代很快，需要按可复现能力和维护状态复核。,ok,,
824874689,cline/cline,"Autonomous coding agent as an SDK, IDE extension, or CLI assistant.",69219,321,7508,1433,Apache-2.0,false,2026-09-24,36.82,33.51,"[202.45,185.78,165.41,158.38,158.92,136.09,114.0,66.44,43.19,47.64,44.01,36.82]","[185.78, 165.41, 158.38, 158.92, 136.09, 114.0, 66.44, 43.19, 47.64, 44.01, 36.82, 33.51]",66,TypeScript,2024-07-06,,keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，65,129 stars，最近可用 OpenRank 49.31。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
846698999,aaif-goose/goose,"an open source, extensible AI agent that goes beyond code suggestions - install, execute, edit, and test with any LLM",54612,452,6308,401,Apache-2.0,false,2026-09-24,21.73,27.75,"[161.49,139.34,136.27,120.49,106.37,107.53,109.64,79.36,66.36,52.99,27.7,21.73]","[139.34, 136.27, 120.49, 106.37, 107.53, 109.64, 79.36, 66.36, 52.99, 27.7, 21.73, 27.75]",21,Rust,2024-08-23,"acp,ai,ai-agents,mcp",keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，51,850 stars，最近可用 OpenRank 30.41。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
937253475,anthropics/claude-code,"Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands.",147919,55,24321,12644,,false,2026-09-24,143.36,76.35,"[677.48,604.1,619.58,558.14,760.1,1059.0,1109.23,1157.77,1069.96,678.34,311.35,143.36]","[604.1, 619.58, 558.14, 760.1, 1059.0, 1109.23, 1157.77, 1069.96, 678.34, 311.35, 143.36, 76.35]",167,TypeScript,2025-02-22,,keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，139,392 stars，最近可用 OpenRank 269.10。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
701547123,open-webui/open-webui,"User-friendly AI Interface (Supports Ollama, OpenAI API, ...)",153022,388,22384,299,NOASSERTION,false,2026-09-24,21.58,33.0,"[237.9,211.8,181.55,159.95,130.05,117.03,128.04,119.73,102.91,83.03,47.32,21.58]","[211.8, 181.55, 159.95, 130.05, 117.03, 128.04, 119.73, 102.91, 83.03, 47.32, 21.58, 33.0]",15,Python,2023-10-06,"ai,llm,llm-ui,llm-webui,llms,mcp,ollama,ollama-webui,open-webui,openai,openapi,rag,self-hosted,ui,webui",keep,Agent Infra,Chatbot workspaces,"当前图中的 Chatbot workspaces 代表项目；仓库仍在维护，147,065 stars，最近可用 OpenRank 43.70。",该区维持三个成熟代表，避免把普通聊天客户端扩张成主图主体。,ok,,
589831718,Comfy-Org/ComfyUI,"The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.",134826,340,15964,4928,GPL-3.0,false,2026-09-24,29.18,24.23,"[158.86,149.9,128.16,130.58,142.74,142.62,117.04,89.44,68.05,61.59,44.57,29.18]","[149.9, 128.16, 130.58, 142.74, 142.62, 117.04, 89.44, 68.05, 61.59, 44.57, 29.18, 24.23]",24,Python,2023-01-17,"ai,comfy,comfyui,python,pytorch,stable-diffusion",remove,Agent Infra,Workflow & agent builders,核心定位是图像和视频生成工作流，不适合作为通用 Workflow & agent builder 代表。,它仍应出现在生成式媒体生态图，而不是本版 Agent Infra 主图。,ok,,
643445235,lobehub/lobehub,"🤯 LobeHub is your Chief Agent Operator, organizing your agents into 7×24 operations by hiring, scheduling, and reporting on your entire AI team.",82800,349,15921,967,NOASSERTION,false,2026-09-24,38.07,36.35,"[111.16,118.49,105.69,91.69,95.44,105.4,94.92,88.57,74.53,65.54,50.82,38.07]","[118.49, 105.69, 91.69, 95.44, 105.4, 94.92, 88.57, 74.53, 65.54, 50.82, 38.07, 36.35]",21,TypeScript,2023-05-21,"agent,agent-collaboration,agent-harness,ai,cao,chatgpt,chief-agent-operator,claude,deepseek,fable,gemini,glm,gpt,knowledge-base,loop-engineering,mcp,openai,skills",keep,Agent Infra,Personal AI assistants,80k+ stars、最近可用 OpenRank 仍在 60 左右，已经成为个人 agent 工作空间和长期运行入口的代表项目。,产品同时覆盖聊天、知识库和多 agent 管理；主图只放在 Personal AI assistants，避免重复出现在 framework 或 chatbot。,ok,,
730534580,infiniflow/ragflow,RAGFlow is a leading open-source Retrieval-Augmented Generation (RAG) engine that fuses cutting-edge RAG with Agent capabilities to create a superior context layer for LLMs,91268,454,10821,1494,Apache-2.0,false,2026-09-24,33.94,44.33,"[199.04,183.85,152.73,151.16,150.85,137.89,107.44,93.33,83.82,76.8,62.05,33.94]","[183.85, 152.73, 151.16, 150.85, 137.89, 107.44, 93.33, 83.82, 76.8, 62.05, 33.94, 44.33]",26,Go,2023-12-12,"agent-harness,agentic-ai,agentic-nagive,agentic-retrieval,agentic-search,ai,ai-agents,context-engine,context-engineering,context-management,harness-engineering,knowledge-compilation,rag,retrieval-augmented-generation,search-harness",keep,Agent Infra,"Memory, knowledge & context","当前图中的 Memory, knowledge & context 代表项目；仓库仍在维护，86,252 stars，最近可用 OpenRank 62.54。",数据库、RAG 和 memory 边界容易混淆，图中要标清每个项目承担的层次。,ok,,
771302083,OpenHands/OpenHands,🙌 OpenHands: AI-Driven Development,89062,460,11725,875,MIT,false,2026-09-24,11.89,18.23,"[114.93,77.2,64.87,50.69,47.45,46.69,41.49,46.02,40.45,24.1,17.37,11.89]","[77.2, 64.87, 50.69, 47.45, 46.69, 41.49, 46.02, 40.45, 24.1, 17.37, 11.89, 18.23]",18,TypeScript,2024-03-13,"agent,artificial-intelligence,chatgpt,claude-ai,cli,developer-tools,gpt,llm,openai",keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，82,399 stars，最近可用 OpenRank 18.42。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
599320067,langflow-ai/langflow,Langflow is a powerful tool for building and deploying AI-powered agents and workflows.,155205,363,10137,1173,MIT,false,2026-09-24,26.11,29.03,"[126.74,121.56,109.35,99.51,95.82,86.53,76.05,69.71,62.27,48.64,36.41,26.11]","[121.56, 109.35, 99.51, 95.82, 86.53, 76.05, 69.71, 62.27, 48.64, 36.41, 26.11, 29.03]",12,Python,2023-02-08,"agents,chatgpt,generative-ai,large-language-models,multiagent,react-flow",keep,Agent Infra,Workflow & agent builders,"当前图中的 Workflow & agent builders 代表项目；仓库仍在维护，152,530 stars，最近可用 OpenRank 43.82。",与通用低代码平台有交叉，主图只保留明确支持 agent workflow 的项目。,ok,,
644461337,vercel/ai,"The AI Toolkit for TypeScript. From the creators of Next.js, the AI SDK is a free open-source library for building AI-powered applications and agents ",26927,470,5184,1477,NOASSERTION,false,2026-09-24,56.02,45.55,"[224.82,205.45,178.68,178.41,174.36,151.8,115.49,117.77,112.85,89.67,69.36,56.02]","[205.45, 178.68, 178.41, 174.36, 151.8, 115.49, 117.77, 112.85, 89.67, 69.36, 56.02, 45.55]",72,TypeScript,2023-05-23,"anthropic,artificial-intelligence,gemini,generative-ai,generative-ui,javascript,language-model,llm,nextjs,openai,react,svelte,typescript,vercel,vue",keep,Agent Infra,Code-first frameworks,"当前图中的 Code-first frameworks 代表项目；仓库仍在维护，25,859 stars，最近可用 OpenRank 55.99。",该区按语言和 runtime 多样性取样，不追求把所有 framework 都放进来。,ok,,
575865240,AstrBotDevs/AstrBot,"AI Agent Assistant & development framework that integrates lots of IM platforms, LLMs, plugins and AI feature, and can be your openclaw alternative. ✨",40960,364,2970,1558,AGPL-3.0,false,2026-09-24,36.97,34.71,"[77.0,77.66,81.73,86.01,82.34,75.71,99.42,108.57,82.91,57.01,45.83,36.97]","[77.66, 81.73, 86.01, 82.34, 75.71, 99.42, 108.57, 82.91, 57.01, 45.83, 36.97, 34.71]",13,Python,2022-12-08,"agent,ai,astrbot,chatbot,chatgpt,discord,docker,gemini,gpt,llama,llm,mcp,openai,python,qq,qqbot,telegram",keep,Agent Infra,Personal AI assistants,"当前图中的 Personal AI assistants 代表项目；仓库仍在维护，38,235 stars，最近可用 OpenRank 42.16。",产品边界普遍较宽，只在这一分区出现，避免与 chatbot 和 framework 重复。,ok,,
818331198,pydantic/pydantic-ai,"How Python does AI. Agents, realtime voice, image generation, embeddings. Every model, every interface, typed end to end.",20146,477,2775,967,MIT,false,2026-09-24,35.72,27.02,"[160.96,130.54,111.76,122.48,133.4,103.31,94.02,88.66,81.01,60.08,40.6,35.72]","[130.54, 111.76, 122.48, 133.4, 103.31, 94.02, 88.66, 81.01, 60.08, 40.6, 35.72, 27.02]",23,Python,2024-06-21,"agent-framework,genai,harness,harness-engineering,llm,pydantic,python",keep,Agent Infra,Code-first frameworks,"当前图中的 Code-first frameworks 代表项目；仓库仍在维护，18,861 stars，最近可用 OpenRank 36.55。",该区按语言和 runtime 多样性取样，不追求把所有 framework 都放进来。,ok,,
656099147,mem0ai/mem0,The Memory Layer for AI Agents - Drop-in memory infrastructure for AI agents and apps. Context that persists. Built for production.,65945,402,7756,755,Apache-2.0,false,2026-09-24,12.93,13.42,"[57.47,49.93,47.76,35.33,24.4,19.14,24.96,25.2,25.05,20.58,18.02,12.93]","[49.93, 47.76, 35.33, 24.4, 19.14, 24.96, 25.2, 25.05, 20.58, 18.02, 12.93, 13.42]",19,Python,2023-06-20,"agentic-memory,agentic-memory-system,agents,ai,ai-agents,chatgpt,genai,llm,long-term-memory,memory,memory-management,python,rag,state-management",keep,Agent Infra,"Memory, knowledge & context","当前图中的 Memory, knowledge & context 代表项目；仓库仍在维护，61,924 stars，最近可用 OpenRank 17.75。",数据库、RAG 和 memory 边界容易混淆，图中要标清每个项目承担的层次。,ok,,
440752086,coder/coder,Secure environments for developers and their agents,16671,268,1583,1107,AGPL-3.0,false,2026-09-24,62.06,47.41,"[90.87,96.98,93.26,78.72,71.37,66.96,71.33,84.79,97.49,90.19,76.5,62.06]","[96.98, 93.26, 78.72, 71.37, 66.96, 71.33, 84.79, 97.49, 90.19, 76.5, 62.06, 47.41]",32,Go,2021-12-22,"agents,dev-tools,development-environment,go,golang,ide,jetbrains,remote-development,terraform,vscode",keep,Agent Infra,Development sandboxes,"当前图中的 Development sandboxes 代表项目；仓库仍在维护，13,970 stars，最近可用 OpenRank 89.97。",编排层和隔离 runtime 不是一回事，主图说明中要保留这一区别。,ok,,
946087422,Kilo-Org/kilocode,"Kilo is the all-in-one agentic engineering platform. Build, ship, and iterate faster with the most popular open source coding agent.",27405,393,3193,612,MIT,false,2026-09-24,33.37,30.79,"[95.43,105.64,104.89,113.88,111.69,110.9,79.87,72.32,71.93,57.72,45.99,33.37]","[105.64, 104.89, 113.88, 111.69, 110.9, 79.87, 72.32, 71.93, 57.72, 45.99, 33.37, 30.79]",23,TypeScript,2025-03-10,"ai,ai-age,ai-coding,ai-developer-tools,chatgpt,claude,cli,gemini,jetbrains,sonnet,vscode,vscode-extension",keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，26,569 stars，最近可用 OpenRank 58.37。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
573661753,activepieces/activepieces,AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents,24710,356,4240,591,NOASSERTION,false,2026-09-24,32.54,31.06,"[86.6,97.57,97.5,87.51,61.6,55.5,75.8,73.79,74.45,66.72,59.73,32.54]","[97.57, 97.5, 87.51, 61.6, 55.5, 75.8, 73.79, 74.45, 66.72, 59.73, 32.54, 31.06]",29,TypeScript,2022-12-03,"ai-agent,ai-agent-tools,ai-agents,ai-agents-framework,mcp,mcp-server,mcp-tools,mcps,n8n-alternative,no-code-automation,workflow,workflow-automation,workflows",keep,Agent Infra,Workflow & agent builders,"当前图中的 Workflow & agent builders 代表项目；仓库仍在维护，23,454 stars，最近可用 OpenRank 61.65。",与通用低代码平台有交叉，主图只保留明确支持 agent workflow 的项目。,ok,,
955620917,upstash/context7,Context7 Platform -- Up-to-date code documentation for LLMs and AI code editors,62392,128,3029,69,MIT,false,2026-09-24,,12.27,"[42.72,39.47,40.13,48.86,46.81,41.89,51.76,42.66,26.71,20.65,13.74,null]","[39.47, 40.13, 48.86, 46.81, 41.89, 51.76, 42.66, 26.71, 20.65, 13.74, null, 12.27]",3,TypeScript,2025-03-26,"llm,mcp,mcp-server,vibe-coding",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；59,901 stars，最近可用 OpenRank 14.38，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
707441527,livekit/agents,A framework for building realtime voice AI agents 🤖🎙️📹 ,14339,459,3792,846,Apache-2.0,false,2026-09-24,21.41,24.81,"[104.31,94.41,89.29,93.27,81.6,71.2,65.35,58.3,48.12,37.95,30.44,21.41]","[94.41, 89.29, 93.27, 81.6, 71.2, 65.35, 58.3, 48.12, 37.95, 30.44, 21.41, 24.81]",10,Python,2023-10-19,"agents,ai,openai,real-time,video,voice",keep,Agent Infra,Code-first frameworks,"当前图中的 Code-first frameworks 代表项目；仓库仍在维护，11,536 stars，最近可用 OpenRank 36.48。",该区按语言和 runtime 多样性取样，不追求把所有 framework 都放进来。,ok,,
958830659,google/adk-python,"An open-source, code-first Python toolkit for building, evaluating, and deploying sophisticated AI agents with flexibility and control.",21626,419,4065,500,Apache-2.0,false,2026-09-24,11.68,16.1,"[118.05,96.93,90.7,103.47,94.9,81.61,68.83,55.68,51.79,35.96,24.77,11.68]","[96.93, 90.7, 103.47, 94.9, 81.61, 68.83, 55.68, 51.79, 35.96, 24.77, 11.68, 16.1]",17,Python,2025-04-01,"agent,agentic,agentic-ai,agents,agents-sdk,ai,ai-agents,aiagentframework,genai,genai-chatbot,llm,llms,multi-agent,multi-agent-systems,multi-agents,multi-agents-collaboration",keep,Agent Infra,Code-first frameworks,"当前图中的 Code-first frameworks 代表项目；仓库仍在维护，20,921 stars，最近可用 OpenRank 25.21。",该区按语言和 runtime 多样性取样，不追求把所有 framework 都放进来。,ok,,
678526156,marimo-team/marimo,"A reactive notebook for Python — run reproducible experiments, query with SQL, execute as a script, deploy as an app, and version with git. Stored as pure Python. All in a modern, AI-native editor.",22889,341,1291,600,Apache-2.0,false,2026-09-24,26.83,22.27,"[88.24,86.25,78.34,69.71,64.46,68.53,66.04,54.21,50.68,45.95,39.2,26.83]","[86.25, 78.34, 69.71, 64.46, 68.53, 66.04, 54.21, 50.68, 45.95, 39.2, 26.83, 22.27]",19,Python,2023-08-14,"artificial-intelligence,dag,data-science,data-visualization,dataflow,developer-tools,machine-learning,notebooks,pipeline,python,reactive,sql,web-app",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；22,094 stars，最近可用 OpenRank 37.27，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
644686905,continuedev/continue,open-source coding agent,36012,393,5425,976,Apache-2.0,false,2026-09-22,2.83,3.73,"[143.42,133.06,106.89,91.62,82.08,79.87,71.35,61.85,35.61,15.73,6.22,2.83]","[133.06, 106.89, 91.62, 82.08, 79.87, 71.35, 61.85, 35.61, 15.73, 6.22, 2.83, 3.73]",5,TypeScript,2023-05-24,"agent,ai,cli,developer-tools,open-source",remove,Agent Infra,Agentic coding,coding agent 区需要压缩；当前 OpenRank 信号明显低于同区头部项目，结构上也没有新增能力层。,Continue 具有历史影响力，本次下架是版面取舍，不等于项目失去价值。,ok,,
552661142,langchain-ai/langchain,The agent engineering platform.,146992,467,24595,568,MIT,false,2026-09-24,27.87,28.64,"[138.98,119.76,116.67,108.12,107.42,95.83,87.21,77.74,67.66,51.07,36.48,27.87]","[119.76, 116.67, 108.12, 107.42, 95.83, 87.21, 77.74, 67.66, 51.07, 36.48, 27.87, 28.64]",16,Python,2022-10-17,"agents,ai,ai-agents,anthropic,chatgpt,deepagents,enterprise,framework,gemini,generative-ai,langchain,langgraph,llm,multiagent,open-source,openai,pydantic,python,rag,typescript",keep,Agent Infra,Code-first frameworks,"当前图中的 Code-first frameworks 代表项目；仓库仍在维护，142,799 stars，最近可用 OpenRank 21.50。",该区按语言和 runtime 多样性取样，不追求把所有 framework 都放进来。,ok,,
805155266,CherryHQ/cherry-studio,"AI productivity studio with smart chat, autonomous agents, and 300+ assistants. Unified access to frontier LLMs",52131,407,5009,1632,AGPL-3.0,false,2026-09-24,45.74,40.43,"[370.94,322.24,253.1,191.58,181.92,153.66,117.61,108.64,99.72,80.35,60.21,45.74]","[322.24, 253.1, 191.58, 181.92, 153.66, 117.61, 108.64, 99.72, 80.35, 60.21, 45.74, 40.43]",17,TypeScript,2024-05-24,"agent-skills,ai-agent,claude-code,codex,deepseek,hermes-agent,open-code-review,skills,vibe-coding",keep,Agent Infra,Chatbot workspaces,"当前图中的 Chatbot workspaces 代表项目；仓库仍在维护，49,083 stars，最近可用 OpenRank 68.81。",该区维持三个成熟代表，避免把普通聊天客户端扩张成主图主体。,ok,,
1108837393,code-yeongyu/oh-my-openagent,"OmO: Just type ""mass ulw"" keyword with your prompt. Now you are the master of graph engineering.",69364,319,5716,1064,NOASSERTION,false,2026-09-24,23.75,23.81,"[null,null,null,null,27.78,115.71,116.74,89.72,73.96,52.49,43.14,23.75]","[null, null, null, 27.78, 115.71, 116.74, 89.72, 73.96, 52.49, 43.14, 23.75, 23.81]",10,TypeScript,2025-12-03,"ai,ai-agents,anthropic,chatgpt,claude,claude-skills,codex,cursor,gemini,ide,openai,opencode,orchestration,tui,typescript",keep,Agent Infra,Coding workflows & harnesses,"当前图中的 Coding harnesses 代表项目；仓库仍在维护，66,711 stars，最近可用 OpenRank 46.25。",harness 热度容易受短期传播影响，需要持续检查外部贡献和真实复用。,ok,,
488641606,agno-agi/agno,"Build, run, and manage agent platforms.",42330,442,6000,1617,Apache-2.0,false,2026-09-24,33.49,29.45,"[121.55,130.26,113.24,95.5,99.9,92.68,75.42,61.57,59.41,51.14,36.66,33.49]","[130.26, 113.24, 95.5, 99.9, 92.68, 75.42, 61.57, 59.41, 51.14, 36.66, 33.49, 29.45]",24,Python,2022-05-04,"agents,ai,ai-agents,developer-tools,python",add,Agent Infra,Code-first frameworks,OpenRank 33.49、7 月参与者 24；替换 Flue 后不增加 framework 分区密度。,与现有 Python agent frameworks 有交叉，后续继续按独立社区和能力边界复核。,ok,,
748713144,huggingface/lerobot,🤗 LeRobot: Making AI for Robotics more accessible with end-to-end learning,27754,269,5734,954,Apache-2.0,false,2026-09-24,9.76,23.05,"[89.54,85.11,85.22,82.25,72.62,61.31,49.73,40.98,39.05,34.6,14.59,9.76]","[85.11, 85.22, 82.25, 72.62, 61.31, 49.73, 40.98, 39.05, 34.6, 14.59, 9.76, 23.05]",24,Python,2024-01-26,,keep,Model Infra,Pre-Train · Robotics infra,"当前图中的 Pre-Train · Robotics infra 代表项目；仓库仍在维护，26,203 stars，最近可用 OpenRank 17.57。",只放 robotics infra，不把机器人 foundation model 混进来。,ok,,
600596928,danny-avila/LibreChat,"Enhanced ChatGPT Clone: Features Agents, MCP, Skills, DeepSeek, Anthropic, AWS, OpenAI, Responses API, Azure, Groq, o1, GPT-5, Mistral, OpenRouter, Vertex AI, Gemini, Artifacts, AI model switching, message search, Code Interpreter, langchain, DALL-E-3, OpenAPI Actions, Functions, Secure Multi-User Auth, Presets, open-source for self-hosting. Active",44858,408,9194,789,MIT,false,2026-09-24,22.06,22.3,"[108.51,103.02,95.08,83.18,73.77,62.06,52.39,44.67,35.55,27.86,24.2,22.06]","[103.02, 95.08, 83.18, 73.77, 62.06, 52.39, 44.67, 35.55, 27.86, 24.2, 22.06, 22.3]",11,TypeScript,2023-02-12,"ai,anthropic,artifacts,aws,azure,chatgpt,chatgpt-clone,claude,clone,deepseek,gemini,google,gpt-5,librechat,mcp,o1,openai,responses-api,vision,webui",keep,Agent Infra,Chatbot workspaces,"当前图中的 Chatbot workspaces 代表项目；仓库仍在维护，41,382 stars，最近可用 OpenRank 29.09。",该区维持三个成熟代表，避免把普通聊天客户端扩张成主图主体。,ok,,
1048065319,thedotmack/claude-mem,"Persistent Context Across Sessions for Every Agent –  Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More",94601,143,8361,295,Apache-2.0,false,2026-09-24,14.75,16.61,"[null,0.83,4.97,8.14,45.93,48.5,54.14,39.98,39.38,31.89,21.61,14.75]","[0.83, 4.97, 8.14, 45.93, 48.5, 54.14, 39.98, 39.38, 31.89, 21.61, 14.75, 16.61]",9,TypeScript,2025-08-31,"ai,ai-agents,ai-memory,anthropic,artificial-intelligence,chromadb,claude,claude-agent-sdk,claude-agents,claude-code,claude-code-plugin,claude-skills,embeddings,long-term-memory,mem0,memory-engine,openmemory,rag,sqlite,supermemory",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；88,830 stars，最近可用 OpenRank 22.41，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
736272311,pipecat-ai/pipecat,"Open Source framework for voice agents, multimodal apps, and realtime AI. Maintained by Daily and the community.",15825,321,2748,371,BSD-2-Clause,false,2026-09-24,16.15,15.88,"[63.28,61.85,53.89,53.99,48.46,50.7,50.75,46.88,33.99,29.8,21.11,16.15]","[61.85, 53.89, 53.99, 48.46, 50.7, 50.75, 46.88, 33.99, 29.8, 21.11, 16.15, 15.88]",8,Python,2023-12-27,"ai,chatbot-framework,chatbots,real-time,voice,voice-assistant",keep,Agent Infra,Code-first frameworks,"当前图中的 Code-first frameworks 代表项目；仓库仍在维护，13,773 stars，最近可用 OpenRank 25.86。",该区按语言和 runtime 多样性取样，不追求把所有 framework 都放进来。,ok,,
914053359,OpenMind/OM1,Modular AI HAL (Hardware Abstraction Layer) for Robots,2932,122,996,70,MIT,false,2026-09-15,,12.86,"[11.39,12.27,19.5,19.99,19.26,79.37,56.85,33.63,20.61,13.53,15.89,null]","[12.27, 19.5, 19.99, 19.26, 79.37, 56.85, 33.63, 20.61, 13.53, 15.89, null, 12.86]",4,Go,2025-01-08,"llm,multiagent,robotics,ros2,zenoh",keep,Model Infra,Pre-Train · Robotics infra,"当前图中的 Pre-Train · Robotics infra 代表项目；仓库仍在维护，2,890 stars，最近可用 OpenRank 15.46。",只放 robotics infra，不把机器人 foundation model 混进来。,ok,,
987670088,charmbracelet/crush,Glamourous agentic coding for all 💘,28280,154,2284,795,NOASSERTION,false,2026-09-24,21.76,19.28,"[121.07,100.96,79.64,62.12,72.1,75.49,68.88,56.86,43.36,40.35,31.77,21.76]","[100.96, 79.64, 62.12, 72.1, 75.49, 68.88, 56.86, 43.36, 40.35, 31.77, 21.76, 19.28]",4,Go,2025-05-21,"agentic-ai,ai,llms,ravishing",remove,Agent Infra,Agentic coding,终端 coding agent 已有足够代表项目，Crush 的功能增量不足以支撑第十三个 logo。,项目仍活跃；若未来形成独特交互或协作模式，可再进入候选池。,ok,,
717197250,QuantumNous/new-api,"A unified AI model hub for aggregation & distribution. It supports cross-converting various LLMs into OpenAI-compatible, Claude-compatible, or Gemini-compatible formats. A centralized gateway for personal and enterprise model management.",48851,306,11725,1334,AGPL-3.0,false,2026-09-23,10.99,12.97,"[66.93,64.37,53.31,51.28,51.44,52.44,50.14,54.96,56.32,48.07,25.34,10.99]","[64.37, 53.31, 51.28, 51.44, 52.44, 50.14, 54.96, 56.32, 48.07, 25.34, 10.99, 12.97]",15,Go,2023-11-10,"ai-gateway,claude,deepseek,gemini,newapi,openai,rerank",keep,Model Infra,Model API gateways,"当前图中的 Model API gateways 代表项目；仓库仍在维护，43,664 stars，最近可用 OpenRank 26.61。",需要区分 model API、MCP gateway 和 agentic proxy，避免把所有代理工具放在一起。,ok,,
1031912724,farion1231/cc-switch,"A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io",136381,250,9314,2884,MIT,false,2026-09-24,29.32,29.73,"[2.52,16.81,23.28,33.81,43.18,60.17,69.66,84.19,76.25,71.67,46.9,29.32]","[16.81, 23.28, 33.81, 43.18, 60.17, 69.66, 84.19, 76.25, 71.67, 46.9, 29.32, 29.73]",28,Rust,2025-08-04,"ai-tools,claude-code,codex,desktop-app,grok,grokbuild,hermes,hermes-agent,mcp,open-source,openclaw,openclaw-ui,opencode,pi,provider-management,rust,skills,skills-management,tauri,wsl-support",keep,Agent Infra,Coding workflows & harnesses,cc-switch 管理 Claude Code、Codex、OpenCode 等 coding agent 的配置和切换，更接近 harness，而不是模型 API gateway。,项目关注度很高，但产品属性较强；保留一个 logo，不在 gateway 区重复出现。,ok,,
1156956890,zeroclaw-labs/zeroclaw,"Fast, small, and fully autonomous AI personal assistant infrastructure, any OS, any platform — deploy anywhere, swap anything 🦀",32874,422,4951,770,Apache-2.0,false,2026-09-24,41.22,32.21,"[null,null,null,null,null,null,90.61,124.01,96.85,67.14,59.25,41.22]","[null, null, null, null, null, 90.61, 124.01, 96.85, 67.14, 59.25, 41.22, 32.21]",21,Rust,2026-02-13,"agent,agentic,ai,infra,ml,openclaw,os,zeroclaw",keep,Agent Infra,Personal AI assistants,"当前图中的 Personal AI assistants 代表项目；仓库仍在维护，32,421 stars，最近可用 OpenRank 54.22。",产品边界普遍较宽，只在这一分区出现，避免与 chatbot 和 framework 重复。,ok,,
1002125012,BloopAI/vibe-kanban,"Get 10X more out of Claude Code, Codex or any coding agent",28181,64,3023,542,Apache-2.0,false,2026-09-19,,3.74,"[31.41,40.99,47.31,44.35,57.29,120.72,118.19,105.34,69.3,28.84,11.04,null]","[40.99, 47.31, 44.35, 57.29, 120.72, 118.19, 105.34, 69.3, 28.84, 11.04, null, 3.74]",1,Rust,2025-06-14,"agent,ai-agents,kanban,management,task-manager",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；27,568 stars，最近可用 OpenRank 4.74，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
710601088,crewAIInc/crewAI,"Framework for orchestrating role-playing, autonomous AI agents. By fostering collaborative intelligence, CrewAI empowers agents to work together seamlessly, tackling complex tasks.",58981,336,8563,473,MIT,false,2026-09-23,14.35,13.6,"[60.94,55.04,52.06,46.94,46.99,38.83,42.77,41.02,38.93,33.39,20.89,14.35]","[55.04, 52.06, 46.94, 46.99, 38.83, 42.77, 41.02, 38.93, 33.39, 20.89, 14.35, 13.6]",19,Python,2023-10-27,"agents,ai,ai-agents,aiagentframework,llms",keep,Agent Infra,Code-first frameworks,"当前图中的 Code-first frameworks 代表项目；仓库仍在维护，56,275 stars，最近可用 OpenRank 20.22。",该区按语言和 runtime 多样性取样，不追求把所有 framework 都放进来。,ok,,
558188628,higress-group/higress,🤖 AI Gateway | AI Native API Gateway,9453,202,1315,1151,Apache-2.0,false,2026-09-24,5.32,11.85,"[68.37,59.5,62.56,54.5,48.33,51.58,41.48,42.87,39.81,27.17,17.58,5.32]","[59.5, 62.56, 54.5, 48.33, 51.58, 41.48, 42.87, 39.81, 27.17, 17.58, 5.32, 11.85]",7,Go,2022-10-27,"ai-gateway,ai-native,api-gateway,cloud-native,envoy",keep,Model Infra,Model API gateways,"当前图中的 Model API gateways 代表项目；仓库仍在维护，8,960 stars，最近可用 OpenRank 20.27。",需要区分 model API、MCP gateway 和 agentic proxy，避免把所有代理工具放在一起。,ok,,
954873280,a2aproject/A2A,Agent2Agent (A2A) is an open protocol enabling communication and interoperability between opaque agentic applications.,25919,162,2625,262,Apache-2.0,false,2026-09-22,5.11,11.42,"[25.6,26.62,20.83,17.26,18.61,18.92,16.03,15.17,13.58,8.02,6.9,5.11]","[26.62, 20.83, 17.26, 18.61, 18.92, 16.03, 15.17, 13.58, 8.02, 6.9, 5.11, 11.42]",8,Shell,2025-03-25,"a2a,a2a-mcp,a2a-protocol,a2a-server,agents,generative-ai,linux-foundation",keep,Agent Infra,Protocols & interoperability,"当前图中的 Protocols & interoperability 代表项目；仓库仍在维护，25,069 stars，最近可用 OpenRank 6.98。",协议成熟度差异较大，preview 或规范状态需要在图上明确。,ok,,
839037098,mastra-ai/mastra,Mastra is the modern TypeScript framework for AI-powered applications and agents.,28311,466,2850,480,NOASSERTION,false,2026-09-24,54.41,60.78,"[146.06,150.79,151.23,147.51,124.33,123.66,122.99,122.2,113.73,101.33,75.58,54.41]","[150.79, 151.23, 147.51, 124.33, 123.66, 122.99, 122.2, 113.73, 101.33, 75.58, 54.41, 60.78]",34,TypeScript,2024-08-06,"agents,ai,chatbots,evals,javascript,llm,mcp,nextjs,nodejs,reactjs,tts,typescript,workflows",keep,Agent Infra,Multi-agent orchestration,"当前图中的 Multi-agent orchestration 代表项目；仓库仍在维护，26,649 stars，最近可用 OpenRank 92.69。",多 agent 概念变化快，保留项目需要持续显示独立编排能力。,ok,,
753490180,daytonaio/daytona,Daytona is a Secure and Elastic Infrastructure for Running AI-Generated Code,71716,230,5647,456,,false,2026-07-24,,8.54,"[35.19,30.72,34.28,36.26,35.49,39.4,36.66,33.94,37.06,27.42,23.41,null]","[30.72, 34.28, 36.26, 35.49, 39.4, 36.66, 33.94, 37.06, 27.42, 23.41, null, 8.54]",1,,2024-02-06,"agentic-workflow,ai,ai-agents,ai-runtime,ai-sandboxes,code-execution,code-interpreter,developer-tools",keep,Agent Infra,Development sandboxes,"当前图中的 Development sandboxes 代表项目；仓库仍在维护，72,122 stars，最近可用 OpenRank 26.80。",编排层和隔离 runtime 不是一回事，主图说明中要保留这一区别。,ok,,
1074561042,gastownhall/beads,Beads - A memory upgrade for your coding agent,27397,353,1856,1291,MIT,false,2026-09-24,20.36,22.96,"[null,null,22.31,32.61,54.29,64.98,68.6,53.27,39.33,34.25,20.07,20.36]","[null, 22.31, 32.61, 54.29, 64.98, 68.6, 53.27, 39.33, 34.25, 20.07, 20.36, 22.96]",13,Go,2025-10-12,"agents,claude-code,coding",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；25,712 stars，最近可用 OpenRank 21.57，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1012087571,router-for-me/CLIProxyAPI,"Wrap Antigravity, ChatGPT Codex, Claude Code, Grok Build, Muse Code, Davin as an OpenAI/Gemini/Claude/Codex compatible API service, allowing you to enjoy the free Gemini Series, GPT Series, Grok Series, Claude model through API",53113,232,8016,644,MIT,false,2026-09-24,21.34,26.16,"[4.33,10.26,17.09,28.38,55.15,81.85,70.82,72.23,71.45,51.91,29.26,21.34]","[10.26, 17.09, 28.38, 55.15, 81.85, 70.82, 72.23, 71.45, 51.91, 29.26, 21.34, 26.16]",13,Go,2025-07-01,"antigravity,claude-code,cluade,codex,devin,gemini,muse,openai",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；45,328 stars，最近可用 OpenRank 31.46，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1027384981,langchain-ai/deepagents,The batteries-included agent harness.,29743,157,4177,185,MIT,false,2026-09-24,159.3,91.18,"[16.98,19.51,33.34,55.2,51.11,99.78,115.95,146.88,154.79,152.18,168.35,159.3]","[19.51, 33.34, 55.2, 51.11, 99.78, 115.95, 146.88, 154.79, 152.18, 168.35, 159.3, 91.18]",62,Python,2025-07-27,"ai,deepagents,harness,harness-engineering,langchain,langgraph,python,typescript",omit,,,Deep Agents 是 LangChain 谱系内的子项目；主图已经保留 LangChain，不重复增加 logo。,可以在 LangChain 的项目演化说明中出现。,ok,,
1132431999,aden-hive/hive,Multi-Agent Harness for Production AI,11069,210,5657,1355,Apache-2.0,false,2026-09-14,,6.93,"[null,null,null,null,null,182.36,257.27,176.81,100.17,37.87,14.71,null]","[null, null, null, null, 182.36, 257.27, 176.81, 100.17, 37.87, 14.71, null, 6.93]",2,Python,2026-01-12,"agent,agent-framework,agent-skills,anthropic,automation,autonomous-agents,claude,harness,harness-engineering,human-in-the-loop,openai,python,self-hosted,self-improving",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；10,782 stars，最近可用 OpenRank 14.14，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1076863218,MoonshotAI/kimi-cli,"[Archived] Legacy Python Kimi CLI, no longer maintained. Please use Kimi Code CLI: https://github.com/MoonshotAI/kimi-code",11430,71,1340,797,Apache-2.0,true,2026-09-22,,4.84,"[null,null,18.31,27.8,22.37,27.86,40.77,33.97,39.08,18.98,9.57,null]","[null, 18.31, 27.8, 22.37, 27.86, 40.77, 33.97, 39.08, 18.98, 9.57, null, 4.84]",1,Python,2025-10-15,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；10,983 stars，最近可用 OpenRank 10.89，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
676672661,langchain-ai/langgraph,Build resilient agents.,42227,280,7147,818,MIT,false,2026-09-23,7.89,6.35,"[97.26,69.88,49.92,49.51,34.35,26.56,25.68,30.53,28.69,22.45,12.2,7.89]","[69.88, 49.92, 49.51, 34.35, 26.56, 25.68, 30.53, 28.69, 22.45, 12.2, 7.89, 6.35]",11,Python,2023-08-09,"agents,ai,ai-agents,chatgpt,deepagents,enterprise,framework,gemini,generative-ai,langchain,langgraph,llm,multiagent,open-source,openai,pydantic,python,rag",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；38,340 stars，最近可用 OpenRank 10.41，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
659402878,spring-projects/spring-ai,An Application Framework for AI Engineering,9480,410,2926,1492,Apache-2.0,false,2026-09-24,12.48,11.38,"[90.58,81.98,71.21,67.56,60.6,51.76,41.05,39.12,27.34,29.89,23.72,12.48]","[81.98, 71.21, 67.56, 60.6, 51.76, 41.05, 39.12, 27.34, 29.89, 23.72, 12.48, 11.38]",7,Java,2023-06-27,"artificial-intelligence,java,spring-ai",omit,,,JVM 生态本版优先加入 Koog；Code-first frameworks 已达到十个项目。,Spring AI 保留为下一轮替换候选。,ok,,
560704231,run-llama/llama_index,LlamaIndex is the document processing platform for AI,52309,476,8206,854,MIT,false,2026-09-24,8.54,9.3,"[82.17,69.79,59.1,43.39,35.86,38.64,37.38,33.17,21.13,19.58,10.83,8.54]","[69.79, 59.1, 43.39, 35.86, 38.64, 37.38, 33.17, 21.13, 19.58, 10.83, 8.54, 9.3]",18,Python,2022-11-02,"agents,application,data,fine-tuning,framework,llamaindex,llm,multi-agents,rag,vector-database",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；51,162 stars，最近可用 OpenRank 9.35，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
607289185,microsoft/semantic-kernel,Integrate cutting-edge LLM technology quickly and easily into your apps,28597,402,4788,321,MIT,false,2026-09-19,1.86,5.99,"[76.45,62.63,32.9,19.33,9.91,11.48,13.42,12.26,8.94,4.18,3.9,1.86]","[62.63, 32.9, 19.33, 9.91, 11.48, 13.42, 12.26, 8.94, 4.18, 3.9, 1.86, 5.99]",6,C#,2023-02-27,"ai,artificial-intelligence,llm,openai,sdk",omit,,,Microsoft Agent Framework 已覆盖其主要 agent framework 位置，避免同一厂商同一谱系重复占位。,Semantic Kernel 继续作为历史和企业集成参考。,ok,,
680120071,microsoft/autogen,A programming framework for agentic AI,61142,443,9249,1098,CC-BY-4.0,false,2026-04-15,4.78,6.1,"[41.88,29.64,12.97,7.48,4.85,5.19,5.42,7.04,5.78,5.85,4.71,4.78]","[29.64, 12.97, 7.48, 4.85, 5.19, 5.42, 7.04, 5.78, 5.85, 4.71, 4.78, 6.1]",9,Python,2023-08-18,"agentic,agentic-agi,agents,ai,autogen,autogen-ecosystem,chatgpt,framework,llm-agent,llm-framework",omit,,,Microsoft Agent Framework 已成为统一入口，本版不再同时保留 AutoGen。,AutoGen 的历史影响力可在框架谱系说明中呈现。,ok,,
214587193,supabase/supabase,"The Postgres development platform. Supabase gives you a dedicated Postgres database to build your web, mobile, and AI applications.",110726,398,14674,1090,Apache-2.0,false,2026-09-24,296.4,160.33,"[209.17,220.03,236.75,260.96,243.22,243.0,253.18,258.4,270.05,287.5,328.1,296.4]","[220.03, 236.75, 260.96, 243.22, 243.0, 253.18, 258.4, 270.05, 287.5, 328.1, 296.4, 160.33]",236,TypeScript,2019-10-12,"ai,alternative,auth,database,deno,embeddings,example,firebase,nextjs,oauth2,pgvector,postgis,postgres,postgresql,postgrest,realtime,supabase,vectors,websockets",remove,Agent Infra,"Memory, knowledge & context",通用后端平台不能直接代表 agent memory；当前 memory 区需要让位给 agent-native context 项目。,Supabase 仍是重要基础设施，但不在本图的有限 logo 预算内。,ok,,
691347156,1Panel-dev/MaxKB,🔥 MaxKB is an open-source platform for building enterprise-grade agents.  强大易用的开源企业级智能体平台。,22875,90,3162,21,GPL-3.0,false,2026-09-24,7.9,11.08,"[69.47,63.52,53.5,46.16,45.34,44.63,39.51,32.83,35.68,32.07,18.15,7.9]","[63.52, 53.5, 46.16, 45.34, 44.63, 39.51, 32.83, 35.68, 32.07, 18.15, 7.9, 11.08]",3,Python,2023-09-14,"agent,agentic-ai,chatbot,deepseek-r1,knowledgebase,langchain,llama3,llm,maxkb,mcp-server,ollama,pgvector,qwen3,rag",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；22,264 stars，最近可用 OpenRank 15.90，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
605673387,labring/FastGPT,"FastGPT is a knowledge-based platform built on the LLMs, offers a comprehensive suite of out-of-the-box capabilities such as data processing, RAG retrieval, and visual AI workflow orchestration, letting you easily develop and deploy complex question-answering systems without the need for extensive setup or configuration.",29739,183,7325,173,NOASSERTION,false,2026-09-24,8.83,15.66,"[65.17,58.51,40.87,40.35,37.05,34.81,26.74,27.45,24.35,18.17,12.37,8.83]","[58.51, 40.87, 40.35, 37.05, 34.81, 26.74, 27.45, 24.35, 18.17, 12.37, 8.83, 15.66]",4,TypeScript,2023-02-23,"agent,claude,deepseek,llm,mcp,nextjs,openai,qwen,rag,workflow",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；29,172 stars，最近可用 OpenRank 14.64，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
621803253,FlowiseAI/Flowise,"Build AI Agents, Visually",55475,317,25049,1040,NOASSERTION,true,2026-08-13,7.76,2.5,"[65.01,49.65,43.2,36.9,33.56,28.59,30.82,29.47,29.14,22.82,12.12,7.76]","[49.65, 43.2, 36.9, 33.56, 28.59, 30.82, 29.47, 29.14, 22.82, 12.12, 7.76, 2.5]",3,TypeScript,2023-03-31,"agentic-ai,agentic-workflow,agents,artificial-intelligence,chatbot,chatgpt,javascript,langchain,large-language-models,low-code,multiagent-systems,no-code,openai,rag,react,typescript,workflow-automation",add,Agent Infra,Workflow & agent builders,54k+ stars 的可视化 agent builder，结构定位比 ComfyUI 更贴合通用 agent workflow。,与 Dify、Langflow 功能接近；该区维持五个 logo，不继续扩容。,ok,,
842586148,yetone/avante.nvim,Use your Neovim like using Cursor AI IDE!,18172,280,850,48,Apache-2.0,false,2026-09-24,3.24,,"[62.36,47.97,30.48,21.69,17.73,32.59,23.62,12.33,7.08,9.15,7.67,3.24]","[47.97, 30.48, 21.69, 17.73, 32.59, 23.62, 12.33, 7.08, 9.15, 7.67, 3.24, null]",2,Lua,2024-08-14,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；18,071 stars，最近可用 OpenRank 6.96，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
23496542,triton-lang/triton,Development repository for the Triton language and compiler,20236,412,3216,1267,MIT,false,2026-09-24,41.95,49.38,"[139.39,137.95,131.66,117.88,110.35,104.34,99.43,88.2,77.45,69.4,47.71,41.95]","[137.95, 131.66, 117.88, 110.35, 104.34, 99.43, 88.2, 77.45, 69.4, 47.71, 41.95, 49.38]",14,MLIR,2014-08-30,,keep,Model Infra,Pre-Train · Compiler & accelerator,"当前图中的 Pre-Train · Compiler & accelerator 代表项目；仓库仍在维护，19,803 stars，最近可用 OpenRank 47.07。",该区强调编译、kernel 和模型优化的结构覆盖，不追求完整 CUDA 工具清单。,ok,,
634043200,modular/modular,The Modular Platform (includes MAX & Mojo),29875,389,3182,1167,NOASSERTION,false,2026-09-23,,25.95,"[89.98,79.21,74.04,69.23,66.24,67.03,67.9,59.3,54.01,41.08,31.65,null]","[79.21, 74.04, 69.23, 66.24, 67.03, 67.9, 59.3, 54.01, 41.08, 31.65, null, 25.95]",3,Mojo,2023-04-28,"ai,language,machine-learning,max,modular,mojo,programming-language",omit,,,编译与 runtime 信号很强，但 GitHub API 未识别标准 SPDX 许可证，本版暂不进入开放生态主图。,完成代码范围和许可证复核后再决定是否替换 compiler 区项目。,ok,,
90506918,NVIDIA/cudf,cuDF - GPU DataFrame Library ,9761,305,1119,1359,Apache-2.0,false,2026-09-24,42.42,37.95,"[96.11,87.37,76.39,70.95,69.4,74.4,69.37,68.5,61.45,63.67,50.08,42.42]","[87.37, 76.39, 70.95, 69.4, 74.4, 69.37, 68.5, 61.45, 63.67, 50.08, 42.42, 37.95]",24,C++,2017-05-07,"arrow,cpp,cuda,cudf,dask,data-analysis,data-science,dataframe,gpu,pandas,pydata,python,rapids",remove,Model Infra,Pre-Train · Compiler & accelerator,cuDF 是 GPU dataframe library，放在 Compiler & accelerator 区的解释成本过高。,它仍是 GPU 数据处理基础设施，但不属于本图的模型编译主线。,ok,,
539057023,NVIDIA/TransformerEngine,"A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit and 4-bit floating point (FP8 and FP4) precision on Hopper, Ada and Blackwell GPUs, to provide better performance with lower memory utilization in both training and inference.",3553,209,835,357,Apache-2.0,false,2026-09-24,17.7,34.61,"[66.75,60.75,52.47,55.65,53.26,46.6,44.06,41.91,46.14,40.7,32.71,17.7]","[60.75, 52.47, 55.65, 53.26, 46.6, 44.06, 41.91, 46.14, 40.7, 32.71, 17.7, 34.61]",9,Python,2022-09-20,"cuda,deep-learning,fp4,fp8,gpu,jax,machine-learning,python,pytorch",keep,Model Infra,Pre-Train · Compiler & accelerator,"当前图中的 Pre-Train · Compiler & accelerator 代表项目；仓库仍在维护，3,450 stars，最近可用 OpenRank 32.73。",该区强调编译、kernel 和模型优化的结构覆盖，不追求完整 CUDA 工具清单。,ok,,
669331976,flashinfer-ai/flashinfer,FlashInfer: Kernel Library for LLM Serving,6502,408,1496,1042,Apache-2.0,false,2026-09-24,147.83,79.8,"[65.19,70.26,78.03,73.14,65.84,71.18,85.64,111.35,127.11,133.35,152.97,147.83]","[70.26, 78.03, 73.14, 65.84, 71.18, 85.64, 111.35, 127.11, 133.35, 152.97, 147.83, 79.8]",119,Cuda,2023-07-22,"attention,cuda,distributed-inference,gpu,jit,large-large-models,llm-inference,moe,nvidia,pytorch",keep,Model Infra,Pre-Train · Compiler & accelerator,"当前图中的 Pre-Train · Compiler & accelerator 代表项目；仓库仍在维护，6,053 stars，最近可用 OpenRank 119.80。",该区强调编译、kernel 和模型优化的结构覆盖，不追求完整 CUDA 工具清单。,ok,,
112542515,NVIDIA/cutlass,CUDA Templates and Python DSLs for High-Performance Linear Algebra,10482,280,2100,756,NOASSERTION,false,2026-09-23,6.04,11.26,"[37.82,35.53,25.17,29.65,27.47,26.48,24.03,20.88,17.89,16.22,10.01,6.04]","[35.53, 25.17, 29.65, 27.47, 26.48, 24.03, 20.88, 17.89, 16.22, 10.01, 6.04, 11.26]",6,C++,2017-11-30,"cpp,cuda,deep-learning,deep-learning-library,gpu,nvidia,python",keep,Model Infra,Pre-Train · Compiler & accelerator,"当前图中的 Pre-Train · Compiler & accelerator 代表项目；仓库仍在维护，10,151 stars，最近可用 OpenRank 6.92。",该区强调编译、kernel 和模型优化的结构覆盖，不追求完整 CUDA 工具清单。,ok,,
724854743,ml-explore/mlx,MLX: An array framework for Apple silicon,28540,335,2276,174,MIT,false,2026-09-24,5.5,13.38,"[40.17,36.07,33.34,34.0,35.75,30.6,30.2,22.56,21.05,18.02,15.68,5.5]","[36.07, 33.34, 34.0, 35.75, 30.6, 30.2, 22.56, 21.05, 18.02, 15.68, 5.5, 13.38]",4,C++,2023-11-28,mlx,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；27,733 stars，最近可用 OpenRank 16.43，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
933928679,deepseek-ai/DeepEP,DeepEP: an efficient expert-parallel communication library,10201,48,1450,289,MIT,false,2026-09-23,,5.25,"[28.29,25.61,19.33,14.46,11.38,8.6,5.56,3.3,5.05,5.38,2.38,null]","[25.61, 19.33, 14.46, 11.38, 8.6, 5.56, 3.3, 5.05, 5.38, 2.38, null, 5.25]",0,Cuda,2025-02-17,,keep,Model Infra,Pre-Train · Compiler & accelerator,"当前图中的 Pre-Train · Compiler & accelerator 代表项目；仓库仍在维护，9,913 stars，最近可用 OpenRank 2.55。",该区强调编译、kernel 和模型优化的结构覆盖，不追求完整 CUDA 工具清单。,ok,,
494232964,Dao-AILab/flash-attention,Fast and memory-efficient exact attention,25007,207,3099,1311,BSD-3-Clause,false,2026-09-24,5.43,5.94,"[30.65,30.05,27.77,23.85,29.74,32.3,30.41,23.43,23.79,13.63,10.88,5.43]","[30.05, 27.77, 23.85, 29.74, 32.3, 30.41, 23.43, 23.79, 13.63, 10.88, 5.43, 5.94]",5,Python,2022-05-19,,keep,Model Infra,Pre-Train · Compiler & accelerator,"当前图中的 Pre-Train · Compiler & accelerator 代表项目；仓库仍在维护，24,557 stars，最近可用 OpenRank 10.59。",该区强调编译、kernel 和模型优化的结构覆盖，不追求完整 CUDA 工具清单。,ok,,
204086862,streamlit/streamlit,Streamlit — A faster way to build and share data apps.,45830,319,4392,1180,Apache-2.0,false,2026-09-24,20.0,20.54,"[101.98,92.94,78.35,75.8,74.68,65.25,66.16,53.29,44.48,33.66,25.72,20.0]","[92.94, 78.35, 75.8, 74.68, 65.25, 66.16, 53.29, 44.48, 33.66, 25.72, 20.0, 20.54]",13,Python,2019-08-24,"data-analysis,data-science,data-visualization,deep-learning,developer-tools,machine-learning,python,streamlit",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；45,375 stars，最近可用 OpenRank 28.95，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
162405963,gradio-app/gradio,"Build and share delightful machine learning apps, all in Python. 🌟 Star to support our work!",43615,466,3610,102,Apache-2.0,false,2026-09-24,11.92,18.1,"[69.7,63.46,52.46,56.73,53.4,46.11,38.05,34.2,31.32,42.09,24.39,11.92]","[63.46, 52.46, 56.73, 53.4, 46.11, 38.05, 34.2, 31.32, 42.09, 24.39, 11.92, 18.1]",3,Python,2018-12-19,"data-analysis,data-science,data-visualization,deep-learning,deploy,gradio,gradio-interface,interface,machine-learning,models,python,python-notebook,ui,ui-components",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；43,220 stars，最近可用 OpenRank 23.38，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
291438522,siyuan-note/siyuan,"An open-source, privacy-first, self-hosted knowledge workspace where humans and AI agents work together 开源、隐私优先、自托管的知识工作空间，让人与智能体在此协作",46488,125,3023,13,AGPL-3.0,false,2026-09-24,23.03,18.04,"[66.78,66.51,63.23,56.79,62.09,60.43,50.04,41.45,38.03,28.39,24.34,23.03]","[66.51, 63.23, 56.79, 62.09, 60.43, 50.04, 41.45, 38.03, 28.39, 24.34, 23.03, 18.04]",12,TypeScript,2020-08-30,"agentic-ai,ai-agent,knowledge-base,local-first,markdown,mcp,note-taking,self-hosted,siyuan",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；45,468 stars，最近可用 OpenRank 29.42，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
826168160,docling-project/docling,Get your documents ready for gen AI,67831,321,4917,926,MIT,false,2026-09-24,40.22,23.99,"[64.23,72.55,80.54,74.66,59.27,48.57,41.62,52.4,59.36,55.5,84.04,40.22]","[72.55, 80.54, 74.66, 59.27, 48.57, 41.62, 52.4, 59.36, 55.5, 84.04, 40.22, 23.99]",14,Python,2024-07-09,"ai,convert,document-parser,document-parsing,documents,docx,html,markdown,pdf,pdf-converter,pdf-to-json,pdf-to-text,pptx,tables,xlsx",add,Model Infra,Data · Integration,文档解析和结构化已经成为 RAG 与 agent 数据准备的常见入口，当前 63k+ stars、最近可用 OpenRank 66.74。,它代表文档数据准备，不是通用 ETL；Data · Integration 区控制在三个项目。,ok,,
649170660,Mintplex-Labs/anything-llm,Stop renting your intelligence. Own it with AnythingLLM. Everything you need for a powerful local-first agent experience ,66406,243,7396,339,MIT,false,2026-09-24,17.92,20.73,"[57.88,51.48,48.16,40.57,35.26,32.53,28.43,28.59,28.3,23.27,19.05,17.92]","[51.48, 48.16, 40.57, 35.26, 32.53, 28.43, 28.59, 28.3, 23.27, 19.05, 17.92, 20.73]",6,JavaScript,2023-06-04,"agent-computer,agent-harness,agent-orchestration,agentic-ai,agentic-workflow,ai-agents,computer-use,hermes-agent,llm,local-ai,localai,multimodal,rag,self-hosted-ai,vector-database",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；64,012 stars，最近可用 OpenRank 18.36，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
158256479,apache/iceberg,Apache Iceberg,9270,407,3550,924,Apache-2.0,false,2026-09-23,34.89,38.82,"[134.55,137.46,112.48,112.97,102.13,98.82,96.01,89.69,83.35,71.45,53.24,34.89]","[137.46, 112.48, 112.97, 102.13, 98.82, 96.01, 89.69, 83.35, 71.45, 53.24, 34.89, 38.82]",32,Java,2018-11-19,"apache,hacktoberfest,iceberg",keep,Model Infra,Data · Governance,"当前图中的 Data · Governance 代表项目；仓库仍在维护，9,085 stars，最近可用 OpenRank 57.16。",Apache 和开放数据格式是该区重点，当前七个项目已经较为饱满。,ok,,
391740664,open-metadata/OpenMetadata,"The Open Context Layer for Data and AI ,  OpenMetadata is the open platform for building trusted data context and business semantics for humans, AI assistants, and agents.",15319,426,2405,899,Apache-2.0,false,2026-09-24,47.49,54.79,"[113.11,116.57,106.35,98.11,92.51,99.44,91.19,89.11,109.97,83.41,61.78,47.49]","[116.57, 106.35, 98.11, 92.51, 99.44, 91.19, 89.11, 109.97, 83.41, 61.78, 47.49, 54.79]",26,TypeScript,2021-08-01,"context,context-layer,data-catalog,data-collaboration,data-contracts,data-discovery,data-governance,data-lineage,data-observability,data-profiling,data-quality,datadiscovery,dataquality,mcp,mcp-server,metadata,metadata-management,ontologies,ontologies-api,semantics",keep,Model Infra,Data · Governance,"当前图中的 Data · Governance 代表项目；仓库仍在维护，14,587 stars，最近可用 OpenRank 76.75。",Apache 和开放数据格式是该区重点，当前七个项目已经较为饱满。,ok,,
46398090,datahub-project/datahub,The Context Platform for your Data and AI Stack,12757,410,3699,1252,Apache-2.0,false,2026-09-24,38.95,43.59,"[100.54,99.22,91.13,91.59,96.72,96.27,88.47,87.19,80.23,69.79,46.2,38.95]","[99.22, 91.13, 91.59, 96.72, 96.27, 88.47, 87.19, 80.23, 69.79, 46.2, 38.95, 43.59]",23,Python,2015-11-18,"agent-platform,context-management,data-catalog,data-discovery,data-governance,data-observability,datahub,metadata",keep,Model Infra,Data · Governance,"当前图中的 Data · Governance 代表项目；仓库仍在维护，12,365 stars，最近可用 OpenRank 54.73。",Apache 和开放数据格式是该区重点，当前七个项目已经较为饱满。,ok,,
631431061,apache/gravitino,"World's most powerful open data catalog for building a high-performance, geo-distributed and federated metadata lake.",3238,343,943,1103,Apache-2.0,false,2026-09-24,22.03,21.93,"[90.47,76.23,61.49,64.26,54.9,47.46,44.34,44.66,37.95,31.45,32.76,22.03]","[76.23, 61.49, 64.26, 54.9, 47.46, 44.34, 44.66, 37.95, 31.45, 32.76, 22.03, 21.93]",13,Java,2023-04-23,"ai-catalog,data-catalog,datalake,federated-query,lakehouse,metadata,metalake,model-catalog,opendatacatalog,skycomputing,stratosphere",keep,Model Infra,Data · Governance,"当前图中的 Data · Governance 代表项目；仓库仍在维护，3,136 stars，最近可用 OpenRank 40.68。",Apache 和开放数据格式是该区重点，当前七个项目已经较为饱满。,ok,,
182849188,delta-io/delta,"An open-source storage framework that enables building a Lakehouse architecture with compute engines including Spark, PrestoDB, Flink, Trino, and Hive and APIs",9021,390,2183,976,Apache-2.0,false,2026-09-24,15.67,27.06,"[61.79,59.29,46.89,44.16,50.32,49.74,43.05,34.88,35.05,35.16,31.22,15.67]","[59.29, 46.89, 44.16, 50.32, 49.74, 43.05, 34.88, 35.05, 35.16, 31.22, 15.67, 27.06]",13,Scala,2019-04-22,"acid,analytics,big-data,delta-lake,spark",keep,Model Infra,Data · Governance,"当前图中的 Data · Governance 代表项目；仓库仍在维护，8,925 stars，最近可用 OpenRank 35.05。",Apache 和开放数据格式是该区重点，当前七个项目已经较为饱满。,ok,,
447060287,apache/paimon,Apache Paimon is a lake format that enables building a Realtime Lakehouse Architecture with Flink and Spark for both streaming and batch operations.,3407,390,1403,750,Apache-2.0,false,2026-09-24,16.38,23.32,"[62.18,57.31,49.55,48.58,50.77,46.53,38.1,34.23,25.2,25.08,15.64,16.38]","[57.31, 49.55, 48.58, 50.77, 46.53, 38.1, 34.23, 25.2, 25.08, 15.64, 16.38, 23.32]",5,Java,2022-01-12,"big-data,data-ingestion,flink,paimon,real-time-analytics,spark,streaming-datalake,table-store",keep,Model Infra,Data · Governance,"当前图中的 Data · Governance 代表项目；仓库仍在维护，3,353 stars，最近可用 OpenRank 18.40。",Apache 和开放数据格式是该区重点，当前七个项目已经较为饱满。,ok,,
76474200,apache/hudi,"Upserts, Deletes And Incremental Processing on Big Data.",6273,384,2523,2904,Apache-2.0,false,2026-09-24,24.44,24.15,"[63.92,65.9,59.1,55.19,58.86,58.51,54.84,48.86,43.18,39.23,31.01,24.44]","[65.9, 59.1, 55.19, 58.86, 58.51, 54.84, 48.86, 43.18, 39.23, 31.01, 24.44, 24.15]",22,Java,2016-12-14,"apacheflink,apachehudi,apachespark,bigdata,data-integration,datalake,hudi,incremental-processing,stream-processing",keep,Model Infra,Data · Governance,"当前图中的 Data · Governance 代表项目；仓库仍在维护，6,197 stars，最近可用 OpenRank 38.83。",Apache 和开放数据格式是该区重点，当前七个项目已经较为饱满。,ok,,
33884891,apache/airflow,"Apache Airflow - A platform to programmatically author, schedule, and monitor workflows",46961,410,17911,2078,Apache-2.0,false,2026-09-24,98.32,81.24,"[334.33,327.95,299.72,261.63,253.36,262.28,271.61,241.59,214.56,168.07,127.22,98.32]","[327.95, 299.72, 261.63, 253.36, 262.28, 271.61, 241.59, 214.56, 168.07, 127.22, 98.32, 81.24]",60,Python,2015-04-13,"airflow,apache,apache-airflow,automation,dag,data-engineering,data-integration,data-orchestrator,data-pipelines,data-science,elt,etl,machine-learning,mlops,orchestration,python,scheduler,workflow,workflow-engine,workflow-orchestration",keep,Model Infra,Data · Integration,"当前图中的 Data · Integration 代表项目；仓库仍在维护，46,289 stars，最近可用 OpenRank 144.80。",只放与 AI 数据准备直接相关的代表项目，避免扩张成通用数据工程图。,ok,,
283046497,airbytehq/airbyte,"Open-source data movement for ELT pipelines and AI agents — from APIs, databases & files to warehouses, lakes, and AI applications. Both self-hosted and Cloud.",22133,378,5360,2498,NOASSERTION,false,2026-09-24,20.61,22.3,"[148.09,138.4,121.97,107.79,86.83,86.97,69.26,69.02,65.19,52.69,33.66,20.61]","[138.4, 121.97, 107.79, 86.83, 86.97, 69.26, 69.02, 65.19, 52.69, 33.66, 20.61, 22.3]",38,Python,2020-07-27,"bigquery,change-data-capture,data,data-analysis,data-collection,data-engineering,data-integration,data-pipeline,elt,etl,java,mssql,mysql,pipeline,postgresql,python,redshift,s3,self-hosted,snowflake",keep,Model Infra,Data · Integration,"当前图中的 Data · Integration 代表项目；仓库仍在维护，21,723 stars，最近可用 OpenRank 43.07。",只放与 AI 数据准备直接相关的代表项目，避免扩张成通用数据工程图。,ok,,
131619646,dagster-io/dagster,"An orchestration platform for the development, production, and observation of data assets.",16199,407,2304,2572,Apache-2.0,false,2026-09-24,16.22,7.84,"[137.42,128.73,115.25,103.57,92.57,87.28,65.78,57.58,40.49,43.1,42.8,16.22]","[128.73, 115.25, 103.57, 92.57, 87.28, 65.78, 57.58, 40.49, 43.1, 42.8, 16.22, 7.84]",17,Python,2018-04-30,"analytics,dagster,data-engineering,data-integration,data-orchestrator,data-pipelines,data-science,etl,metadata,mlops,orchestration,python,scheduler,workflow,workflow-automation",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；15,910 stars，最近可用 OpenRank 25.79，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
192640529,HumanSignal/label-studio,Label Studio is a multi-type data labeling and annotation tool with standardized output format,28341,149,3729,949,Apache-2.0,false,2026-09-24,1.97,1.75,"[67.06,63.58,61.59,55.91,50.78,50.66,44.36,38.1,16.13,7.16,null,1.97]","[63.58, 61.59, 55.91, 50.78, 50.66, 44.36, 38.1, 16.13, 7.16, null, 1.97, 1.75]",4,TypeScript,2019-06-19,"annotation,annotation-tool,annotations,boundingbox,computer-vision,data-labeling,dataset,datasets,deep-learning,image-annotation,image-classification,image-labeling,image-labelling-tool,label-studio,labeling,labeling-tool,mlops,semantic-segmentation,text-annotation,yolo",keep,Model Infra,Data · Labeling,"当前图中的 Data · Labeling 代表项目；仓库仍在维护，27,939 stars，最近窗口暂未产出 OpenRank。",保留通用标注平台，垂直数据工具进入专题而不是主图。,ok,,
139156354,cvat-ai/cvat,"Computer Vision Annotation Tool (CVAT) is a leading platform for building high-quality visual datasets for vision AI. It offers open-source, cloud, and enterprise products, as well as labeling services, for image, video, and 3D annotation with AI-assisted labeling, quality assurance, team collaboration, analytics, and developer APIs.",16785,284,3885,585,MIT,false,2026-09-24,6.34,18.82,"[52.58,44.78,42.19,38.39,38.77,32.21,31.51,26.45,25.72,23.91,11.47,6.34]","[44.78, 42.19, 38.39, 38.77, 32.21, 31.51, 26.45, 25.72, 23.91, 11.47, 6.34, 18.82]",5,Python,2018-06-29,"annotation,annotation-tool,annotations,boundingbox,computer-vision,computer-vision-annotation,dataset,deep-learning,image-annotation,image-classification,image-labeling,image-labelling-tool,imagenet,labeling,labeling-tool,object-detection,pytorch,semantic-segmentation,tensorflow,video-annotation",keep,Model Infra,Data · Labeling,"当前图中的 Data · Labeling 代表项目；仓库仍在维护，16,398 stars，最近可用 OpenRank 13.74。",保留通用标注平台，垂直数据工具进入专题而不是主图。,ok,,
60377070,vespa-engine/vespa,The AI search platform,7109,117,738,253,Apache-2.0,false,2026-09-24,,17.72,"[42.12,38.62,36.55,34.99,37.19,34.17,40.67,35.66,33.19,32.61,12.87,null]","[38.62, 36.55, 34.99, 37.19, 34.17, 40.67, 35.66, 33.19, 32.61, 12.87, null, 17.72]",1,Java,2016-06-03,"ai,big-data,java,machine-learning,rag,search,search-engine,server,serving-recommendation,tensor,vector,vector-database,vector-search,vespa",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；7,034 stars，最近可用 OpenRank 15.26，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
71932349,ray-project/ray,Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.,43915,410,8080,3621,Apache-2.0,false,2026-09-24,71.45,66.47,"[275.21,273.44,270.3,243.95,228.69,205.21,189.76,179.88,167.56,139.69,101.64,71.45]","[273.44, 270.3, 243.95, 228.69, 205.21, 189.76, 179.88, 167.56, 139.69, 101.64, 71.45, 66.47]",42,Python,2016-10-25,"data-science,deep-learning,deployment,distributed,hyperparameter-optimization,hyperparameter-search,large-language-models,llm,llm-inference,llm-serving,machine-learning,optimization,parallel,python,pytorch,ray,reinforcement-learning,rllib,serving,tensorflow",keep,Model Infra,Compute & scheduling,"当前图中的 Compute & scheduling 代表项目；仓库仍在维护，43,374 stars，最近可用 OpenRank 117.28。",保留通用计算和 Kubernetes AI 调度代表，不增加相似控制器。,ok,,
17165658,apache/spark,Apache Spark - A unified analytics engine for large-scale data processing,44036,331,29399,578,Apache-2.0,false,2026-09-24,68.18,58.43,"[193.41,177.0,162.9,166.92,157.89,136.64,141.5,129.54,116.31,94.85,84.69,68.18]","[177.0, 162.9, 166.92, 157.89, 136.64, 141.5, 129.54, 116.31, 94.85, 84.69, 68.18, 58.43]",44,Scala,2014-02-25,"big-data,java,jdbc,python,r,scala,spark,sql",keep,Model Infra,Compute & scheduling,"当前图中的 Compute & scheduling 代表项目；仓库仍在维护，43,716 stars，最近可用 OpenRank 87.87。",保留通用计算和 Kubernetes AI 调度代表，不增加相似控制器。,ok,,
175592968,volcano-sh/volcano,A Cloud Native Batch System (Project under CNCF),5979,329,1548,832,Apache-2.0,false,2026-09-24,9.67,11.01,"[36.71,35.04,25.85,18.66,28.12,29.18,22.7,22.7,21.46,13.32,13.04,9.67]","[35.04, 25.85, 18.66, 28.12, 29.18, 22.7, 22.7, 21.46, 13.32, 13.04, 9.67, 11.01]",12,Go,2019-03-14,"ai,batch-systems,bigdata,gene,golang,hpc,kubernetes,machine-learning,serving,training",keep,Model Infra,Compute & scheduling,"当前图中的 Compute & scheduling 代表项目；仓库仍在维护，5,813 stars，最近可用 OpenRank 15.19。",保留通用计算和 Kubernetes AI 调度代表，不增加相似控制器。,ok,,
200722670,NVIDIA-NeMo/Speech,"A scalable generative AI framework built for researchers and developers working on Large Language Models, Multimodal, and Speech AI (Automatic Speech Recognition and Text-to-Speech)",18501,400,3613,300,Apache-2.0,false,2026-09-23,10.26,12.53,"[124.39,98.25,68.53,56.51,52.17,45.1,38.39,37.3,29.47,17.45,19.52,10.26]","[98.25, 68.53, 56.51, 52.17, 45.1, 38.39, 37.3, 29.47, 17.45, 19.52, 10.26, 12.53]",7,Python,2019-08-05,"asr,deeplearning,generative-ai,machine-translation,neural-networks,speaker-diariazation,speaker-recognition,speech-synthesis,speech-translation,tts",remove,Model Infra,Pre-Train · Framework & parallel,原 NeMo 仓库已经改名为 Speech，继续放在通用 Pre-Train framework 会造成事实错误。,Speech 项目可在语音模型或多模态工具图中单独评估。,ok,,
235860204,deepspeedai/DeepSpeed,"DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.",43158,435,5005,1468,Apache-2.0,false,2026-09-24,5.74,15.23,"[45.71,44.34,34.6,28.89,23.88,22.9,18.28,19.31,17.31,10.36,7.86,5.74]","[44.34, 34.6, 28.89, 23.88, 22.9, 18.28, 19.31, 17.31, 10.36, 7.86, 5.74, 15.23]",5,Python,2020-01-23,"billion-parameters,compression,data-parallelism,deep-learning,gpu,inference,machine-learning,mixture-of-experts,model-parallelism,pipeline-parallelism,pytorch,trillion-parameters,zero",keep,Model Infra,Pre-Train · Framework & parallel,"当前图中的 Pre-Train · Framework & parallel 代表项目；仓库仍在维护，42,826 stars，最近可用 OpenRank 6.65。",保留通用框架和并行训练基线，不按短期 stars 排名。,ok,,
176982014,NVIDIA/Megatron-LM,Ongoing research training transformer models at scale,17999,339,4562,1409,NOASSERTION,false,2026-09-24,32.39,39.04,"[25.27,24.28,37.21,58.69,74.89,87.23,88.23,83.3,68.93,61.62,47.44,32.39]","[24.28, 37.21, 58.69, 74.89, 87.23, 88.23, 83.3, 68.93, 61.62, 47.44, 32.39, 39.04]",36,Python,2019-03-21,"large-language-models,model-para,transformers",keep,Model Infra,Pre-Train · Framework & parallel,"当前图中的 Pre-Train · Framework & parallel 代表项目；仓库仍在维护，17,247 stars，最近可用 OpenRank 54.06。",保留通用框架和并行训练基线，不按短期 stars 排名。,ok,,
850267588,78/xiaozhi-esp32,An MCP-based chatbot | 一个基于MCP的聊天机器人,30186,152,7042,669,MIT,false,2026-09-23,5.51,8.35,"[57.16,48.48,39.48,38.44,37.04,30.32,24.93,17.13,14.25,8.02,6.34,5.51]","[48.48, 39.48, 38.44, 37.04, 30.32, 24.93, 17.13, 14.25, 8.02, 6.34, 5.51, 8.35]",2,C++,2024-08-31,"chatbot,esp32,mcp",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；28,429 stars，最近可用 OpenRank 6.18，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
712225112,Genesis-Embodied-AI/genesis-world,Simulation platform for general-purpose robotics & embodied AI learning.,29981,110,2867,129,Apache-2.0,false,2026-09-24,8.46,12.88,"[56.49,42.44,37.61,31.52,23.29,31.82,29.68,21.16,11.04,6.38,8.09,8.46]","[42.44, 37.61, 31.52, 23.29, 31.82, 29.68, 21.16, 11.04, 6.38, 8.09, 8.46, 12.88]",3,Python,2023-10-31,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；29,656 stars，最近可用 OpenRank 8.22，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
673412924,modelscope/ms-swift,"Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.8, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM5.3, Gemma4, Llava, Phi4, ...) (AAAI 2025).",15723,224,1702,577,Apache-2.0,false,2026-09-24,13.69,14.63,"[142.65,144.62,123.77,117.91,104.07,97.84,67.11,57.88,37.95,24.22,18.68,13.69]","[144.62, 123.77, 117.91, 104.07, 97.84, 67.11, 57.88, 37.95, 24.22, 18.68, 13.69, 14.63]",10,Python,2023-08-01,"deepseek,embedding,glm,grpo,internvl,kimi,liger,llama,llm,lora,megatron,moe,multimodal,peft,qwen,qwen3,qwen3-omni,qwen3-vl,reranker,sft",keep,Model Infra,Post-Train · Supervised fine-tuning,"当前图中的 Post-Train · Supervised fine-tuning 代表项目；仓库仍在维护，14,973 stars，最近可用 OpenRank 22.32。",当前三个代表已经覆盖主流易用训练工具，不继续扩容。,ok,,
725205304,unslothai/unsloth,"Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4, MiniMax-H3, Gemma 4, FLUX and more.",76701,335,7025,1292,Apache-2.0,false,2026-09-24,28.52,28.53,"[85.45,72.9,63.93,59.17,62.49,53.3,38.65,49.81,45.06,40.08,38.53,28.52]","[72.9, 63.93, 59.17, 62.49, 53.3, 38.65, 49.81, 45.06, 40.08, 38.53, 28.52, 28.53]",13,Python,2023-11-29,"agent,ai,chatgpt,deepseek,fine-tuning,gemma,image-generation,llama,llm,llms,openai,python,qwen,reinforcement-learning,self-hosted,stable-diffusion,text-to-speech,tts,ui,unsloth",keep,Model Infra,Post-Train · Supervised fine-tuning,"当前图中的 Post-Train · Supervised fine-tuning 代表项目；仓库仍在维护，69,015 stars，最近可用 OpenRank 42.76。",当前三个代表已经覆盖主流易用训练工具，不继续扩容。,ok,,
646410686,hiyouga/LlamaFactory,Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024),75003,282,9188,1157,Apache-2.0,false,2026-09-14,,7.71,"[105.06,79.56,55.45,44.04,40.39,36.8,25.72,19.31,16.38,14.31,5.48,null]","[79.56, 55.45, 44.04, 40.39, 36.8, 25.72, 19.31, 16.38, 14.31, 5.48, null, 7.71]",1,Python,2023-05-28,"agent,ai,deepseek,fine-tuning,gemma,gpt,instruction-tuning,large-language-models,llama,llama3,llm,lora,moe,nlp,peft,qlora,quantization,qwen,rlhf,transformers",keep,Model Infra,Post-Train · Supervised fine-tuning,"当前图中的 Post-Train · Supervised fine-tuning 代表项目；仓库仍在维护，73,582 stars，最近可用 OpenRank 6.19。",当前三个代表已经覆盖主流易用训练工具，不继续扩容。,ok,,
658928958,ollama/ollama,"Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.",181602,451,17990,4066,MIT,false,2026-09-24,167.01,75.39,"[305.53,241.51,239.99,230.8,208.74,202.57,216.92,242.09,270.02,230.61,248.76,167.01]","[241.51, 239.99, 230.8, 208.74, 202.57, 216.92, 242.09, 270.02, 230.61, 248.76, 167.01, 75.39]",152,Go,2023-06-26,"deepseek,gemma,gemma3,glm,go,golang,gpt-oss,llama,llama3,llm,llms,minimax,mistral,ollama,qwen",keep,Model Infra,Serving · Deploy,"当前图中的 Serving · Deploy 代表项目；仓库仍在维护，177,103 stars，最近可用 OpenRank 187.55。",该区只保留部署入口和 serving control plane，不与底层 inference engine 重复。,ok,,
942206898,ai-dynamo/dynamo,A Datacenter Scale Distributed Inference Serving Framework,8155,405,1617,1564,NOASSERTION,false,2026-09-24,115.92,95.31,"[151.15,151.5,153.69,153.06,150.3,152.61,159.09,171.21,173.74,167.43,141.32,115.92]","[151.5, 153.69, 153.06, 150.3, 152.61, 159.09, 171.21, 173.74, 167.43, 141.32, 115.92, 95.31]",62,Rust,2025-03-03,"diffusion,disaggregated-serving,kubernetes,llm-inference,omni,routing-engine,rust,sglang,tensorrt-llm,vllm",keep,Model Infra,Serving · Deploy,"当前图中的 Serving · Deploy 代表项目；仓库仍在维护，7,608 stars，最近可用 OpenRank 170.51。",该区只保留部署入口和 serving control plane，不与底层 inference engine 重复。,ok,,
653496050,xorbitsai/inference,"Swap GPT for any LLM by changing a single line of code. Xinference lets you run open-source, speech, and multimodal models on cloud, on-prem, or your laptop — all through one unified, production-ready inference API.",9591,168,867,34,Apache-2.0,false,2026-09-23,6.32,10.29,"[62.59,45.38,37.24,34.59,29.45,27.79,25.69,15.16,14.44,9.58,7.84,6.32]","[45.38, 37.24, 34.59, 29.45, 27.79, 25.69, 15.16, 14.44, 9.58, 7.84, 6.32, 10.29]",6,Python,2023-06-14,"artificial-intelligence,deployment,diffusers,gemma,glm,glm-5-3,inference,kimi,kimi-k3,llama-cpp,llamacpp,llm,machine-learning,openai-api,pytorch,qwen,sglang,transformers,vllm,whisper",remove,Model Infra,Serving · Deploy,Serving · Deploy 已有 Ollama、Dynamo 和 llm-d，Xorbits Inference 的近期社区信号相对较弱。,项目仍有部署价值；本次下架主要是控制 serving 区密度。,ok,,
833306239,containers/ramalama,"RamaLama is an open-source developer tool that simplifies the local serving of AI models from any source and facilitates their use for inference in production, all through the familiar language of containers.",3059,137,371,117,MIT,false,2026-09-24,5.22,6.95,"[43.02,37.44,32.24,28.43,24.94,22.68,18.93,17.79,17.45,12.44,9.04,5.22]","[37.44, 32.24, 28.43, 24.94, 22.68, 18.93, 17.79, 17.45, 12.44, 9.04, 5.22, 6.95]",1,Python,2024-07-24,"ai,containers,cuda,hacktoberfest,hip,inference-server,intel,llamacpp,llm,podman,vllm",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；2,976 stars，最近可用 OpenRank 8.27，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
799049436,gpustack/gpustack,"A GPU cluster manager for high-performance AI model serving (vLLM, SGLang) and on-demand SSH-accessible GPU instances.",5755,56,661,707,Apache-2.0,false,2026-09-24,14.44,13.32,"[49.02,40.31,36.85,43.63,56.02,51.16,41.18,43.32,32.02,24.09,21.31,14.44]","[40.31, 36.85, 43.63, 56.02, 51.16, 41.18, 43.32, 32.02, 24.09, 21.31, 14.44, 13.32]",5,Python,2024-05-11,"ascend,cuda,deepseek,distributed-inference,genai,high-performance-inference,inference,llama,llm,llm-inference,llm-serving,maas,mindie,openai,qwen,rocm,sglang,vllm",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；5,397 stars，最近可用 OpenRank 26.62，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
599547518,vllm-project/vllm,A high-throughput and memory-efficient inference and serving engine for LLMs,92623,452,22615,8380,Apache-2.0,false,2026-09-24,189.08,148.64,"[762.07,764.83,697.99,674.73,620.83,586.12,560.08,507.78,446.03,360.39,274.25,189.08]","[764.83, 697.99, 674.73, 620.83, 586.12, 560.08, 507.78, 446.03, 360.39, 274.25, 189.08, 148.64]",164,Python,2023-02-09,"amd,blackwell,cuda,deepseek,deepseek-v3,gpt,gpt-oss,inference,kimi,llama,llm,llm-serving,model-serving,moe,openai,pytorch,qwen,qwen3,tpu,transformer",keep,Model Infra,Serving · Inference,"当前图中的 Serving · Inference 代表项目；仓库仍在维护，87,453 stars，最近可用 OpenRank 307.32。",项目数量已经较多，子项目和硬件 plugin 优先作为注释而不是独立 logo。,ok,,
740303686,sgl-project/sglang,SGLang is a high-performance serving framework for large language models and multimodal models.,36403,450,9121,5409,Apache-2.0,false,2026-09-24,215.0,160.3,"[406.94,434.55,415.14,440.36,463.56,469.22,425.68,392.12,360.89,322.31,263.19,215.0]","[434.55, 415.14, 440.36, 463.56, 469.22, 425.68, 392.12, 360.89, 322.31, 263.19, 215.0, 160.3]",112,Python,2024-01-08,"attention,blackwell,cuda,deepseek,diffusion,glm,gpt-oss,inference,llama,llm,minimax,moe,qwen,qwen-image,reinforcement-learning,transformer,vlm,wan",keep,Model Infra,Serving · Inference,"当前图中的 Serving · Inference 代表项目；仓库仍在维护，30,852 stars，最近可用 OpenRank 314.95。",项目数量已经较多，子项目和硬件 plugin 优先作为注释而不是独立 logo。,ok,,
679366051,NVIDIA/TensorRT-LLM,TensorRT LLM provides users with an easy-to-use Python API to define Large Language Models (LLMs) and supports state-of-the-art optimizations to perform inference efficiently on NVIDIA GPUs. TensorRT LLM also contains components to create Python and C++ runtimes that orchestrate the inference execution in a performant way.,14710,421,2771,1502,NOASSERTION,false,2026-09-24,145.76,138.69,"[314.49,309.07,295.11,293.49,298.25,300.45,281.85,276.53,272.23,266.5,214.2,145.76]","[309.07, 295.11, 293.49, 298.25, 300.45, 281.85, 276.53, 272.23, 266.5, 214.2, 145.76, 138.69]",77,Python,2023-08-16,"blackwell,cuda,llm-serving,moe,pytorch",keep,Model Infra,Serving · Inference,"当前图中的 Serving · Inference 代表项目；仓库仍在维护，14,235 stars，最近可用 OpenRank 215.38。",项目数量已经较多，子项目和硬件 plugin 优先作为注释而不是独立 logo。,ok,,
612354784,ggml-org/llama.cpp,LLM inference in C/C++,129423,447,23685,2532,MIT,false,2026-09-24,90.46,80.9,"[293.89,268.53,242.87,264.39,263.88,243.8,234.98,224.73,216.77,186.81,142.46,90.46]","[268.53, 242.87, 264.39, 263.88, 243.8, 234.98, 224.73, 216.77, 186.81, 142.46, 90.46, 80.9]",53,C++,2023-03-10,ggml,keep,Model Infra,Serving · Inference,"当前图中的 Serving · Inference 代表项目；仓库仍在维护，121,872 stars，最近可用 OpenRank 158.32。",项目数量已经较多，子项目和硬件 plugin 优先作为注释而不是独立 logo。,ok,,
153097643,openvinotoolkit/openvino,OpenVINO™ is an open source toolkit for optimizing and deploying AI inference,10910,409,3403,879,Apache-2.0,false,2026-09-24,27.34,35.69,"[203.95,193.62,172.78,157.73,144.15,159.96,139.38,126.15,109.86,74.42,44.65,27.34]","[193.62, 172.78, 157.73, 144.15, 159.96, 139.38, 126.15, 109.86, 74.42, 44.65, 27.34, 35.69]",39,C++,2018-10-15,"ai,computer-vision,deep-learning,deploy-ai,diffusion-models,generative-ai,good-first-issue,inference,llm-inference,natural-language-processing,nlp,openvino,optimize-ai,performance-boost,recommendation-system,speech-recognition,stable-diffusion,transformers,yolo",keep,Model Infra,Serving · Inference,"当前图中的 Serving · Inference 代表项目；仓库仍在维护，10,585 stars，最近可用 OpenRank 53.67。",项目数量已经较多，子项目和硬件 plugin 优先作为注释而不是独立 logo。,ok,,
136202695,mlflow/mlflow,"The open source AI engineering platform for agents, LLMs, and ML models. MLflow enables teams of all sizes to debug, evaluate, monitor, and optimize production-quality AI applications while controlling costs and managing access to models and data.",28125,446,6352,2144,Apache-2.0,false,2026-09-24,31.5,28.1,"[111.37,107.38,106.57,105.73,107.73,111.34,106.97,96.34,79.88,66.96,48.66,31.5]","[107.38, 106.57, 105.73, 107.73, 111.34, 106.97, 96.34, 79.88, 66.96, 48.66, 31.5, 28.1]",22,Python,2018-06-05,"agentops,agents,ai,ai-governance,apache-spark,evaluation,langchain,llm-evaluation,llmops,machine-learning,ml,mlflow,mlops,model-management,observability,open-source,openai,prompt-engineering",keep,Model Infra,Pre-Train · Evaluation & observability,"当前图中的 Pre-Train · Evaluation & observability 代表项目；仓库仍在维护，27,247 stars，最近可用 OpenRank 52.06。",该区同时包含实验追踪和模型生命周期管理，需避免与 Agent observability 混用。,ok,,
515647260,1Panel-dev/1Panel,"🔥 1Panel is a modern, open-source Linux server management panel and a lightweight AI management platform.",37014,124,3360,334,GPL-3.0,false,2026-09-24,14.66,9.15,"[105.75,91.18,73.81,59.74,71.2,63.36,49.25,42.27,29.97,22.23,14.04,14.66]","[91.18, 73.81, 59.74, 71.2, 63.36, 49.25, 42.27, 29.97, 22.23, 14.04, 14.66, 9.15]",8,Go,2022-07-19,"agent,deepseek-harness,docker,docker-ui,hermes,hermes-agent,linux,lnmp,ollama,openclaw,openresty,qwenpaw,webmin",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；36,337 stars，最近可用 OpenRank 14.00，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
642497346,langfuse/langfuse,"🪢 Open source agent evals & observability: Trace, evaluate, and improve LLM applications with one open platform.",35008,209,3842,927,NOASSERTION,false,2026-09-24,31.4,34.77,"[100.04,104.09,99.96,97.3,91.02,77.48,72.29,64.31,58.68,46.11,41.18,31.4]","[104.09, 99.96, 97.3, 91.02, 77.48, 72.29, 64.31, 58.68, 46.11, 41.18, 31.4, 34.77]",22,TypeScript,2023-05-18,"analytics,autogen,evaluation,langchain,large-language-models,llama-index,llm,llm-evaluation,llm-observability,llmops,monitoring,observability,open-source,openai,playground,prompt-engineering,prompt-management,self-hosted,ycombinator",keep,Agent Infra,Observability & evaluation,"当前图中的 Observability & evaluation 代表项目；仓库仍在维护，32,022 stars，最近可用 OpenRank 38.80。",商业产品和 open-core 较多，后续要持续检查许可证与开放治理。,ok,,
86031674,wandb/wandb,"The AI developer platform. Use Weights & Biases to train and fine-tune models, and manage models from experimentation to production.",11260,232,899,987,MIT,false,2026-09-24,33.87,28.99,"[84.7,86.37,77.47,76.89,71.21,67.76,63.6,59.54,51.61,43.27,39.45,33.87]","[86.37, 77.47, 76.89, 71.21, 67.76, 63.6, 59.54, 51.61, 43.27, 39.45, 33.87, 28.99]",8,Python,2017-03-24,"ai,collaboration,data-science,data-versioning,deep-learning,experiment-track,hyperparameter-optimization,hyperparameter-search,hyperparameter-tuning,jax,keras,machine-learning,ml-platform,mlops,model-versioning,pytorch,reinforcement-learning,reproducibility,tensorflow",keep,Model Infra,Pre-Train · Evaluation & observability,"当前图中的 Pre-Train · Evaluation & observability 代表项目；仓库仍在维护，11,209 stars，最近可用 OpenRank 33.98。",该区同时包含实验追踪和模型生命周期管理，需避免与 Agent observability 混用。,ok,,
638951438,comet-ml/opik,"Debug, evaluate, and monitor your LLM applications, RAG systems, and agentic workflows with comprehensive tracing, automated evaluations, and production-ready dashboards.",22218,153,1826,246,Apache-2.0,false,2026-09-24,27.22,28.4,"[68.95,68.86,70.39,70.94,70.1,74.01,66.56,63.06,59.82,45.0,35.93,27.22]","[68.86, 70.39, 70.94, 70.1, 74.01, 66.56, 63.06, 59.82, 45.0, 35.93, 27.22, 28.4]",20,Python,2023-05-10,"evaluation,hacktoberfest,hacktoberfest2025,langchain,llama-index,llm,llm-evaluation,llm-observability,llmops,open-source,openai,playground,prompt-engineering",keep,Agent Infra,Observability & evaluation,"当前图中的 Observability & evaluation 代表项目；仓库仍在维护，20,941 stars，最近可用 OpenRank 45.14。",商业产品和 open-core 较多，后续要持续检查许可证与开放治理。,ok,,
564072810,Arize-ai/phoenix,AI Observability & Evaluation,11600,230,1153,1032,NOASSERTION,false,2026-09-24,25.11,25.64,"[67.03,70.12,61.89,58.86,52.71,47.7,44.34,42.22,38.49,35.44,29.59,25.11]","[70.12, 61.89, 58.86, 52.71, 47.7, 44.34, 42.22, 38.49, 35.44, 29.59, 25.11, 25.64]",15,Python,2022-11-09,"agents,ai-monitoring,ai-observability,aiengineering,anthropic,datasets,evals,langchain,llamaindex,llm-eval,llm-evaluation,llmops,llms,openai,prompt-engineering,smolagents",keep,Agent Infra,Observability & evaluation,"当前图中的 Observability & evaluation 代表项目；仓库仍在维护，10,780 stars，最近可用 OpenRank 37.35。",商业产品和 open-core 较多，后续要持续检查许可证与开放治理。,ok,,
205706595,mlrun/mlrun,"MLRun is an open source MLOps platform for quickly building and managing continuous ML applications across their lifecycle. MLRun integrates into your development and CI/CD environment and automates the delivery of production data, ML pipelines, and online applications.",1697,94,314,119,Apache-2.0,false,2026-09-24,,9.6,"[48.87,44.01,45.74,43.86,39.18,36.45,38.21,25.44,19.49,20.89,13.38,null]","[44.01, 45.74, 43.86, 39.18, 36.45, 38.21, 25.44, 19.49, 20.89, 13.38, null, 9.6]",2,Python,2019-09-01,"data-engineering,data-science,experiment-tracking,kubernetes,machine-learning,mlops,mlops-workflow,model-serving,python,workflow",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；1,688 stars，最近可用 OpenRank 17.65，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
633927609,promptfoo/promptfoo,"Test your prompts, agents, and RAGs. Red teaming/pentesting/vulnerability scanning for AI. Compare performance of GPT, Claude, Gemini, DeepSeek, and more. Simple declarative configs with command line and CI/CD integration.  Used by OpenAI and Anthropic.",25417,345,2366,638,MIT,false,2026-09-24,10.32,10.58,"[43.54,42.89,42.2,43.2,46.16,49.33,47.3,34.79,28.64,24.27,15.69,10.32]","[42.89, 42.2, 43.2, 46.16, 49.33, 47.3, 34.79, 28.64, 24.27, 15.69, 10.32, 10.58]",11,TypeScript,2023-04-28,"ci,ci-cd,cicd,evaluation,evaluation-framework,llm,llm-eval,llm-evaluation,llm-evaluation-framework,llmops,pentesting,prompt-engineering,prompt-testing,prompts,rag,red-teaming,testing,vulnerability-scanners",keep,Agent Infra,Observability & evaluation,"当前图中的 Observability & evaluation 代表项目；仓库仍在维护，23,689 stars，最近可用 OpenRank 17.26。",商业产品和 open-core 较多，后续要持续检查许可证与开放治理。,ok,,
222824662,dagger/dagger,"Automation engine to build, test and ship any codebase. Runs locally, in CI, or directly in the cloud",16298,302,929,195,Apache-2.0,false,2026-09-24,11.51,18.34,"[47.33,50.45,47.68,35.58,32.95,32.61,29.24,28.2,26.87,21.97,19.87,11.51]","[50.45, 47.68, 35.58, 32.95, 32.61, 29.24, 28.2, 26.87, 21.97, 19.87, 11.51, 18.34]",7,Go,2019-11-20,"agents,caching,ci-cd,containers,continuous-deployment,continuous-integration,devops,docker,graphql,workflows",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；16,104 stars，最近可用 OpenRank 22.36，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
615510678,camel-ai/camel,🐫 CAMEL: The first and the best multi-agent framework. Finding the Scaling Law of Agents. https://www.camel-ai.org,17767,233,2091,514,Apache-2.0,false,2026-09-20,5.54,7.41,"[53.32,44.08,36.41,35.95,47.66,32.86,27.08,19.23,12.16,10.0,7.48,5.54]","[44.08, 36.41, 35.95, 47.66, 32.86, 27.08, 19.23, 12.16, 10.0, 7.48, 5.54, 7.41]",2,Python,2023-03-17,"agent,ai-societies,artificial-intelligence,communicative-ai,cooperative-ai,deep-learning,large-language-models,multi-agent-systems,natural-language-processing",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；17,499 stars，最近可用 OpenRank 9.21，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
946380199,openai/openai-agents-python,"A lightweight, powerful framework for multi-agent workflows",29675,385,4802,49,MIT,false,2026-09-23,11.7,13.09,"[94.44,83.2,65.04,54.06,38.26,30.06,24.18,24.24,23.74,22.35,20.64,11.7]","[83.2, 65.04, 54.06, 38.26, 30.06, 24.18, 24.24, 23.74, 22.35, 20.64, 11.7, 13.09]",7,Python,2025-03-11,"agents,ai,framework,harness,llm,openai,python",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；28,238 stars，最近可用 OpenRank 18.53，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
826170402,elizaOS/eliza,Open source agentic operating system,19495,122,5761,1598,MIT,false,2026-09-24,19.63,24.72,"[56.5,45.04,28.69,19.02,16.15,14.33,9.43,7.45,8.43,21.01,13.44,19.63]","[45.04, 28.69, 19.02, 16.15, 14.33, 9.43, 7.45, 8.43, 21.01, 13.44, 19.63, 24.72]",15,TypeScript,2024-07-09,"agent,agentic,ai,autonomous,chatbot,crypto,discord,eliza,elizaos,framework,plugins,rag,slack,swarm,telegram",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；18,821 stars，最近可用 OpenRank 14.26，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
881221486,verl-project/verl,verl/HybridFlow: A Flexible and Efficient RL Post-Training Framework ,23601,445,4588,1257,Apache-2.0,false,2026-09-24,22.94,24.84,"[291.3,258.16,246.46,255.35,251.6,225.13,148.99,106.43,83.06,51.81,33.54,22.94]","[258.16, 246.46, 255.35, 251.6, 225.13, 148.99, 106.43, 83.06, 51.81, 33.54, 22.94, 24.84]",13,Python,2024-10-31,,keep,Model Infra,Post-Train · RL & environments,"当前图中的 Post-Train · Reinforcement learning 代表项目；仓库仍在维护，22,699 stars，最近可用 OpenRank 40.18。",框架功能重叠明显，按通用性和持续协作控制在四个项目。,ok,,
672415139,OpenRLHF/OpenRLHF,"An Easy-to-use, Scalable and High-performance Agentic RL Framework based on Ray (PPO & DAPO & REINFORCE++ &  VLM & TIS & vLLM & Ray & Async  RL)",10042,94,1024,387,Apache-2.0,false,2026-09-17,,1.92,"[17.71,13.98,11.7,7.71,8.54,7.74,6.93,6.39,6.17,5.32,null,null]","[13.98, 11.7, 7.71, 8.54, 7.74, 6.93, 6.39, 6.17, 5.32, null, null, 1.92]",0,Python,2023-07-30,"large-language-models,proximal-policy-optimization,raylib,reinforcement-learning,reinforcement-learning-from-human-feedback,transformers,visual-language-models,vllm",remove,Model Infra,Post-Train · RL & environments,Post-Train 区加入 TRL 后需要一进一出；OpenRLHF 最近窗口缺少可靠 OpenRank。,项目仍是 RLHF 工程实践的重要参考，本次只是不再占主图 logo。,ok,,
937952934,areal-project/AReaL,The RL Bridge for LLM-based Agent Applications. Made Simple & Flexible.,5793,103,608,134,Apache-2.0,false,2026-09-24,6.45,,"[25.34,35.25,42.88,45.74,49.92,43.66,40.36,60.93,57.26,38.83,16.51,6.45]","[35.25, 42.88, 45.74, 49.92, 43.66, 40.36, 60.93, 57.26, 38.83, 16.51, 6.45, null]",4,Python,2025-02-24,"agent,llm,llm-agent,llm-reasoning,machine-learning-systems,mlsys,reinforcement-learning,rl",keep,Model Infra,Post-Train · RL & environments,"当前图中的 Post-Train · Reinforcement learning 代表项目；仓库仍在维护，5,612 stars，最近可用 OpenRank 16.43。",框架功能重叠明显，按通用性和持续协作控制在四个项目。,ok,,
65600975,pytorch/pytorch,Tensors and Dynamic neural networks in Python with strong GPU acceleration,103247,327,30188,17586,NOASSERTION,false,2026-09-24,333.26,253.31,"[988.28,969.04,874.64,868.4,862.71,816.66,743.38,685.38,631.31,537.92,427.62,333.26]","[969.04, 874.64, 868.4, 862.71, 816.66, 743.38, 685.38, 631.31, 537.92, 427.62, 333.26, 253.31]",149,Python,2016-08-13,"autograd,deep-learning,gpu,machine-learning,neural-network,numpy,python,tensor",keep,Model Infra,Pre-Train · Framework & parallel,"当前图中的 Pre-Train · Framework & parallel 代表项目；仓库仍在维护，102,035 stars，最近可用 OpenRank 436.28。",保留通用框架和并行训练基线，不按短期 stars 排名。,ok,,
65711522,PaddlePaddle/Paddle,PArallel Distributed Deep LEarning: Machine Learning Framework from Industrial Practice （『飞桨』核心框架，深度学习&机器学习高性能单机、分布式训练和跨平台部署）,24103,360,6023,1479,Apache-2.0,false,2026-09-24,23.89,18.5,"[186.91,166.93,144.29,140.34,130.77,121.26,110.58,92.03,70.67,62.24,39.17,23.89]","[166.93, 144.29, 140.34, 130.77, 121.26, 110.58, 92.03, 70.67, 62.24, 39.17, 23.89, 18.5]",38,C++,2016-08-15,"deep-learning,distributed-training,efficiency,machine-learning,neural-network,paddlepaddle,python,scalability",keep,Model Infra,Pre-Train · Framework & parallel,"当前图中的 Pre-Train · Framework & parallel 代表项目；仓库仍在维护，24,035 stars，最近可用 OpenRank 43.25。",保留通用框架和并行训练基线，不按短期 stars 排名。,ok,,
154739597,jax-ml/jax,"Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more",36333,396,3804,2584,Apache-2.0,false,2026-09-24,121.48,65.78,"[134.17,118.76,128.69,117.39,112.25,106.2,108.49,114.18,121.46,117.86,135.11,121.48]","[118.76, 128.69, 117.39, 112.25, 106.2, 108.49, 114.18, 121.46, 117.86, 135.11, 121.48, 65.78]",77,Python,2018-10-25,jax,keep,Model Infra,Pre-Train · Framework & parallel,"当前图中的 Pre-Train · Framework & parallel 代表项目；仓库仍在维护，36,065 stars，最近可用 OpenRank 84.71。",保留通用框架和并行训练基线，不按短期 stars 排名。,ok,,
208728772,milvus-io/milvus,"Milvus is a high-performance, cloud-native vector database built for scalable vector ANN search",46247,335,4270,1473,Apache-2.0,false,2026-09-24,44.93,33.11,"[116.51,107.86,102.03,107.23,102.79,98.91,90.06,90.99,89.43,76.66,59.38,44.93]","[107.86, 102.03, 107.23, 102.79, 98.91, 90.06, 90.99, 89.43, 76.66, 59.38, 44.93, 33.11]",30,Go,2019-09-16,"anns,cloud-native,diskann,distributed,embedding-database,embedding-similarity,embedding-store,faiss,golang,hnsw,image-search,llm,nearest-neighbor-search,rag,vector-database,vector-search,vector-similarity,vector-store",add,Agent Infra,"Memory, knowledge & context",成熟向量数据库仍是 agent context 的重要底座；当前 45k+ stars、最近可用 OpenRank 68.34，社区信号稳定。,它是通用向量数据库，不应被描述成 agent memory framework；在图中承担底层检索基础设施的位置。,ok,,
546206616,chroma-core/chroma,Search infrastructure for AI,29365,190,2531,887,Apache-2.0,false,2026-09-24,14.4,17.21,"[63.96,54.0,50.77,50.99,47.47,44.82,41.11,39.41,37.15,23.47,17.18,14.4]","[54.0, 50.77, 50.99, 47.47, 44.82, 41.11, 39.41, 37.15, 23.47, 17.18, 14.4, 17.21]",10,Rust,2022-10-05,"agents,ai,ai-agents,database,rust,rust-lang",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；28,897 stars，最近可用 OpenRank 18.62，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
55072677,weaviate/weaviate,"Weaviate is an open-source vector database that stores both objects and vectors, allowing for the combination of vector search with structured filtering with the fault tolerance and scalability of a cloud-native database​.",16842,159,1407,742,NOASSERTION,false,2026-09-24,26.72,27.27,"[63.36,60.17,55.03,46.03,45.07,45.17,41.06,41.28,33.24,39.43,28.02,26.72]","[60.17, 55.03, 46.03, 45.07, 45.17, 41.06, 41.28, 33.24, 39.43, 28.02, 26.72, 27.27]",17,Go,2016-03-30,"approximate-nearest-neighbor-search,generative-search,grpc,hnsw,hybrid-search,image-search,information-retrieval,mlops,nearest-neighbor-search,neural-search,recommender-system,search-engine,semantic-search,semantic-search-engine,similarity-search,vector-database,vector-search,vector-search-engine,vectors,weaviate",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；16,654 stars，最近可用 OpenRank 32.00，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
268163609,qdrant/qdrant,"Qdrant - High-performance, massive-scale Vector Database and Vector Search Engine for the next generation of AI. Also available in the cloud https://cloud.qdrant.io/",34787,197,2694,751,Apache-2.0,false,2026-09-24,16.61,21.09,"[61.08,59.28,53.52,46.19,47.81,41.79,40.52,43.15,38.95,35.37,25.57,16.61]","[59.28, 53.52, 46.19, 47.81, 41.79, 40.52, 43.15, 38.95, 35.37, 25.57, 16.61, 21.09]",7,Rust,2020-05-30,"ai-search,ai-search-engine,embeddings-similarity,hnsw,hybrid-search,image-search,knn-algorithm,machine-learning,mlops,nearest-neighbor-search,neural-network,neural-search,recommender-system,search,search-engine,search-engines,similarity-search,vector-database,vector-search,vector-search-engine",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；33,626 stars，最近可用 OpenRank 30.03，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
833984877,kvcache-ai/ktransformers,A Flexible Framework for Experiencing Heterogeneous LLM Inference/Fine-tune Optimizations,19536,117,1575,507,Apache-2.0,false,2026-09-23,3.1,5.06,"[18.36,14.14,7.07,16.69,26.48,19.9,17.52,11.75,12.58,6.61,6.24,3.1]","[14.14, 7.07, 16.69, 26.48, 19.9, 17.52, 11.75, 12.58, 6.61, 6.24, 3.1, 5.06]",2,Python,2024-07-26,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；19,081 stars，最近可用 OpenRank 6.96，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
813348592,vllm-project/aibrix,Cost-efficient and pluggable Infrastructure components for GenAI inference,5111,184,699,384,Apache-2.0,false,2026-09-24,8.85,13.11,"[38.95,37.06,33.4,26.81,23.38,19.7,17.26,13.61,14.89,12.71,9.92,8.85]","[37.06, 33.4, 26.81, 23.38, 19.7, 17.26, 13.61, 14.89, 12.71, 9.92, 8.85, 13.11]",3,Go,2024-06-10,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；4,982 stars，最近可用 OpenRank 11.94，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
819733173,kvcache-ai/Mooncake,"Mooncake is the serving platform for Kimi, a leading LLM service provided by Moonshot AI.",6652,345,1266,599,Apache-2.0,false,2026-09-24,20.65,27.01,"[40.1,47.14,38.72,41.69,47.71,51.75,47.99,50.6,43.03,36.4,25.37,20.65]","[47.14, 38.72, 41.69, 47.71, 51.75, 47.99, 50.6, 43.03, 36.4, 25.37, 20.65, 27.01]",25,C++,2024-06-25,"disaggregation,inference,kvcache,llm,rdma,reinforcement-learning,sglang,tokenspeed,trt-llm,vllm",remove,Model Infra,Serving · Inference,KV cache 位置与 LMCache 重叠，本版选择社区信号更强、定位更直接的 LMCache。,Mooncake 继续保留在项目池，后续按跨引擎采用情况复核。,ok,,
814708478,unitycatalog/unitycatalog,"Open, Multi-modal Catalog for Data & AI",3542,130,673,447,Apache-2.0,false,2026-09-22,8.49,12.38,"[4.66,6.33,9.29,9.45,8.54,7.97,10.12,9.21,9.67,9.3,5.43,8.49]","[6.33, 9.29, 9.45, 8.54, 7.97, 10.12, 9.21, 9.67, 9.3, 5.43, 8.49, 12.38]",6,Java,2024-06-13,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；3,472 stars，最近可用 OpenRank 6.80，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
762304524,ComposioHQ/composio,"Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to help you build AI agents that turn intent into action.",30309,103,4820,84,MIT,false,2026-09-24,10.03,10.55,"[27.74,32.0,26.63,25.97,25.72,28.88,28.25,27.45,27.49,22.74,15.2,10.03]","[32.0, 26.63, 25.97, 25.72, 28.88, 28.25, 27.45, 27.49, 22.74, 15.2, 10.03, 10.55]",10,TypeScript,2024-02-23,"agentic-ai,agents,ai,ai-agents,aiagents,developer-tools,function-calling,gpt-4,javascript,js,llm,llmops,mcp,python,remote-mcp-server,sse,typescript",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；29,417 stars，最近可用 OpenRank 15.67，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
839372398,zaidmukaddam/scira,Scira (Formerly MiniPerplx) is a minimalistic AI-powered search engine that helps you find information on the internet and cites it too. Powered by Vercel AI SDK!,11899,20,1500,26,AGPL-3.0,false,2026-08-12,,,"[7.87,2.57,1.53,4.71,3.66,1.62,1.7,2.11,1.13,null,0.43,null]","[2.57, 1.53, 4.71, 3.66, 1.62, 1.7, 2.11, 1.13, null, 0.43, null, null]",0,TypeScript,2024-08-07,"ai-search-engine,minimalistic-ai-search-engine,scira,scira-ai,search-engine",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；11,816 stars，最近可用 OpenRank 0.39，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
357241735,searxng/searxng,SearXNG is a free internet metasearch engine which aggregates results from various search services and databases. Users are neither tracked nor profiled.,37595,277,3421,206,AGPL-3.0,false,2026-09-23,9.0,14.12,"[32.44,33.39,33.25,30.76,31.91,23.81,16.55,20.41,14.48,13.63,13.79,9.0]","[33.39, 33.25, 30.76, 31.91, 23.81, 16.55, 20.41, 14.48, 13.63, 13.79, 9.0, 14.12]",4,Python,2021-04-12,"bing,brave,degoogle,duckduckgo,google,metasearch,privacy,python,qwant,search,search-engine,searx,searxng,startpage,yahoo",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；34,544 stars，最近可用 OpenRank 13.26，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
761150311,Zipstack/unstract,LLM-Driven Extraction of Unstructured Data — Built for API Deployments & ETL Pipeline Workflows,7257,30,720,96,AGPL-3.0,false,2026-09-24,16.3,15.75,"[37.42,39.01,35.86,35.91,34.2,36.78,32.58,35.75,28.35,26.5,20.56,16.3]","[39.01, 35.86, 35.91, 34.2, 36.78, 32.58, 35.75, 28.35, 26.5, 20.56, 16.3, 15.75]",7,Python,2024-02-21,"ai-agents,data-engineering,document-ai,generative-ai,idp,json-extraction,llm,mcp-server,ocr,pdf-extraction,prompt-engineering,structured-output",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；6,940 stars，最近可用 OpenRank 19.65，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
534011733,Farama-Foundation/Gymnasium,"A standard API for single-agent reinforcement learning environments, with popular reference environments and related utilities (formerly Gym)",12576,441,1479,91,MIT,false,2026-09-21,,4.48,"[10.63,13.76,9.59,7.97,7.94,7.49,5.74,6.33,6.59,5.31,3.57,null]","[13.76, 9.59, 7.97, 7.94, 7.49, 5.74, 6.33, 6.59, 5.31, 3.57, null, 4.48]",1,Python,2022-09-08,"api,gym,reinforcement-learning",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；12,238 stars，最近可用 OpenRank 1.51，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
860252770,microsoft/OmniParser,A simple screen parsing tool towards pure vision based GUI agent,25449,10,2235,232,CC-BY-4.0,false,2026-07-20,,,"[5.11,4.87,2.83,1.3,1.01,0.81,null,null,0.37,null,null,null]","[4.87, 2.83, 1.3, 1.01, 0.81, null, null, 0.37, null, null, null, null]",0,Jupyter Notebook,2024-09-20,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；25,196 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
541798154,Unstructured-IO/unstructured,"Convert documents to structured data effortlessly. Unstructured is open-source ETL solution for transforming complex documents into clean, structured formats for language models.  Visit our website to learn more about our enterprise grade Platform product for production grade workflows, partitioning, enrichments, chunking and embedding.",15481,149,1336,313,Apache-2.0,false,2026-09-24,1.95,5.45,"[14.71,9.82,6.67,6.32,4.74,7.52,9.26,5.95,6.36,4.23,3.86,1.95]","[9.82, 6.67, 6.32, 4.74, 7.52, 9.26, 5.95, 6.36, 4.23, 3.86, 1.95, 5.45]",3,HTML,2022-09-26,"data-pipelines,deep-learning,document-image-analysis,document-image-processing,document-parser,document-parsing,docx,donut,information-retrieval,langchain,llm,machine-learning,ml,natural-language-processing,nlp,ocr,pdf,pdf-to-json,pdf-to-text,preprocessing",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；15,210 stars，最近可用 OpenRank 3.27，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
610260322,chatboxai/chatbox,Powerful AI Client,41853,71,4260,1303,GPL-3.0,false,2026-09-24,,4.1,"[27.82,21.13,14.94,12.06,20.45,21.19,12.68,8.06,5.78,3.07,2.05,null]","[21.13, 14.94, 12.06, 20.45, 21.19, 12.68, 8.06, 5.78, 3.07, 2.05, null, 4.1]",0,TypeScript,2023-03-06,"assistant,chatbot,chatgpt,claude,claude-code,copilot,deepseek,gemini,gpt,gpt-5,ollama,openai",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；41,180 stars，最近可用 OpenRank 2.27，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
178976529,bentoml/BentoML,"The easiest way to serve AI apps and models - Build Model Inference APIs, Job queues, LLM apps, Multi-model pipelines, and more!",8858,230,1035,223,Apache-2.0,false,2026-09-07,,4.39,"[11.28,14.96,10.06,5.56,5.23,4.71,2.58,2.68,1.26,0.88,null,null]","[14.96, 10.06, 5.56, 5.23, 4.71, 2.58, 2.68, 1.26, 0.88, null, null, 4.39]",0,Python,2019-04-02,"ai-inference,deep-learning,generative-ai,inference-platform,llm,llm-inference,llm-serving,llmops,machine-learning,ml-engineering,mlops,model-inference-service,model-serving,multimodal,python",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；8,739 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
178075572,kserve/kserve,"Standardized Distributed Generative and Predictive AI Inference Platform for Scalable, Multi-Framework Deployment on Kubernetes",5996,369,1701,217,Apache-2.0,false,2026-09-24,9.58,18.95,"[20.37,17.93,19.47,19.71,15.86,19.09,21.52,18.8,20.61,21.37,20.09,9.58]","[17.93, 19.47, 19.71, 15.86, 19.09, 21.52, 18.8, 20.61, 21.37, 20.09, 9.58, 18.95]",24,Go,2019-03-27,"artificial-intelligence,cncf,genai,hacktoberfest,istio,k8s,knative,kserve,kubeflow,kubernetes,llm-inference,machine-learning,mlops,model-interpretability,model-serving,pytorch,service-mesh,tensorflow,vllm,xgboost",keep,Model Infra,Compute & scheduling,"当前图中的 Compute & scheduling 代表项目；仓库仍在维护，5,740 stars，最近可用 OpenRank 23.47。",保留通用计算和 Kubernetes AI 调度代表，不增加相似控制器。,ok,,
221654678,deepset-ai/haystack,"Open-source AI orchestration framework for building context-engineered, production-ready LLM applications. Design modular pipelines and agent workflows with explicit control over retrieval, routing, memory, and generation. Built for scalable agents, RAG, multimodal applications, semantic search, and conversational systems.",26589,445,3176,155,Apache-2.0,false,2026-09-24,9.95,11.27,"[24.69,21.54,22.99,24.45,24.28,20.88,26.95,23.41,20.14,21.58,16.16,9.95]","[21.54, 22.99, 24.45, 24.28, 20.88, 26.95, 23.41, 20.14, 21.58, 16.16, 9.95, 11.27]",7,Python,2019-11-14,"agent-framework,agentic-ai,agentic-rag,agents,ai,ai-agents,context-engineering,framework,genai,generative-ai,information-retrieval,large-language-models,llm,mcp,multi-agent,orchestration,python,rag,retrieval-augmented-generation,semantic-search",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；26,043 stars，最近可用 OpenRank 18.96，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
820144741,datachain-ai/datachain,"The Context Layer for unstructured data: typed, versioned datasets over S3, GCS, Azure",2817,31,159,101,Apache-2.0,false,2026-09-24,7.53,9.86,"[17.97,17.64,14.29,13.39,13.33,11.14,11.64,11.8,12.02,10.33,9.97,7.53]","[17.64, 14.29, 13.39, 13.33, 11.14, 11.64, 11.8, 12.02, 10.33, 9.97, 7.53, 9.86]",2,Python,2024-06-25,"ai-agents,claude-code,codex,data-context-layer,data-processing,harness-engineering,knowledge-base,mlops,multimodal,pydantic,unstructured-data",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；2,802 stars，最近可用 OpenRank 12.55，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
46153892,NVIDIA/nccl,Optimized primitives for collective multi-GPU communication,5117,133,1426,447,NOASSERTION,false,2026-09-23,,6.46,"[20.74,20.03,13.08,13.19,16.12,14.48,8.12,8.58,8.0,7.53,3.91,null]","[20.03, 13.08, 13.19, 16.12, 14.48, 8.12, 8.58, 8.0, 7.53, 3.91, null, 6.46]",0,C++,2015-11-14,"communications,cpp,cuda,deep-learning,gpu,nvidia",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；4,917 stars，最近可用 OpenRank 4.15，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
614765452,Significant-Gravitas/AutoGPT,"AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.",187521,425,45994,536,NOASSERTION,false,2026-09-24,24.14,24.27,"[45.66,43.19,40.23,39.77,37.42,40.09,41.29,44.96,41.13,37.66,31.98,24.14]","[43.19, 40.23, 39.77, 37.42, 40.09, 41.29, 44.96, 41.13, 37.66, 31.98, 24.14, 24.27]",16,Python,2023-03-16,"agentic-ai,agents,ai,artificial-intelligence,autonomous-agents,claude,gpt,llama-api,llm,openai,python",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；185,734 stars，最近可用 OpenRank 30.10，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
949271792,mindverse/Second-Me,"Train your AI self, amplify you, bridge the world",15687,34,1209,143,Apache-2.0,false,2025-09-30,,,"[8.61,5.21,1.46,1.3,null,0.99,null,null,null,0.44,null,null]","[5.21, 1.46, 1.3, null, 0.99, null, null, null, 0.44, null, null, null]",1,Python,2025-03-16,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；15,628 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
53548867,dbt-labs/dbt-core,dbt enables data analysts and engineers to transform their data using the same practices that software engineers use to build applications.,13920,379,2584,1592,Apache-2.0,false,2026-09-24,14.11,20.15,"[35.98,33.26,26.54,31.18,28.0,32.34,30.84,30.36,25.39,20.69,28.24,14.11]","[33.26, 26.54, 31.18, 28.0, 32.34, 30.84, 30.36, 25.39, 20.69, 28.24, 14.11, 20.15]",13,Rust,2016-03-10,"analytics,business-intelligence,data-modeling,dbt-viewpoint,elt,pypa,slack",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；13,533 stars，最近可用 OpenRank 25.30，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
139199684,PrefectHQ/prefect,Prefect is a workflow orchestration framework for building resilient data pipelines in Python.,23918,423,2538,851,Apache-2.0,false,2026-09-24,10.51,14.43,"[47.65,39.85,33.62,31.15,35.55,35.3,35.53,30.2,22.07,20.46,15.52,10.51]","[39.85, 33.62, 31.15, 35.55, 35.3, 35.53, 30.2, 22.07, 20.46, 15.52, 10.51, 14.43]",7,Python,2018-06-29,"automation,data,data-engineering,data-ops,data-science,infrastructure,ml-ops,observability,orchestration,pipeline,prefect,python,workflow,workflow-engine",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；23,505 stars，最近可用 OpenRank 17.36，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
580642043,oobabooga/textgen,"Open-source desktop app for local LLMs. Text, vision, tool-calling, OpenAI/Anthropic-compatible API. 100% private.",47701,381,5995,842,AGPL-3.0,false,2026-08-17,0.92,1.29,"[22.75,20.63,15.9,8.72,8.76,7.19,3.61,7.54,7.7,5.95,1.73,0.92]","[20.63, 15.9, 8.72, 8.76, 7.19, 3.61, 7.54, 7.7, 5.95, 1.73, 0.92, 1.29]",1,Python,2022-12-21,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；47,503 stars，最近可用 OpenRank 1.97，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
422274596,hpcaitech/ColossalAI,"Making large AI models cheaper, faster and more accessible",41440,190,4498,508,Apache-2.0,false,2026-09-24,,0.38,"[9.28,6.18,null,3.95,2.32,2.03,0.9,null,0.66,null,null,null]","[6.18, null, 3.95, 2.32, 2.03, 0.9, null, 0.66, null, null, null, 0.38]",1,Python,2021-10-28,"ai,big-model,data-parallelism,deep-learning,distributed-computing,foundation-models,heterogeneous-training,hpc,inference,large-scale,model-parallelism,pipeline-parallelism",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；41,423 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
599524116,SillyTavern/SillyTavern,LLM Frontend for Power Users.,33739,385,6351,626,AGPL-3.0,false,2026-09-23,5.42,8.53,"[35.29,30.84,31.7,31.6,30.62,35.07,32.54,29.67,26.7,19.76,9.92,5.42]","[30.84, 31.7, 31.6, 30.62, 35.07, 32.54, 29.67, 26.7, 19.76, 9.92, 5.42, 8.53]",3,JavaScript,2023-02-09,"ai,chat,llm",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；31,266 stars，最近可用 OpenRank 11.89，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
527591471,AUTOMATIC1111/stable-diffusion-webui,Stable Diffusion web UI,165098,429,31377,2507,AGPL-3.0,false,2026-03-02,0.85,1.24,"[9.72,6.17,5.04,6.96,5.21,4.72,6.54,4.38,1.64,1.61,null,0.85]","[6.17, 5.04, 6.96, 5.21, 4.72, 6.54, 4.38, 1.64, 1.61, null, 0.85, 1.24]",2,Python,2022-08-22,"ai,ai-art,deep-learning,diffusion,gradio,image-generation,image2image,img2img,pytorch,stable-diffusion,text2image,torch,txt2img,unstable,upscaling,web",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；164,294 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
28782747,dask/dask,Parallel computing with task scheduling,13924,417,1965,1341,BSD-3-Clause,false,2026-08-24,,5.2,"[21.11,17.09,16.6,21.67,15.7,17.98,13.77,14.37,12.77,10.45,7.36,null]","[17.09, 16.6, 21.67, 15.7, 17.98, 13.77, 14.37, 12.77, 10.45, 7.36, null, 5.2]",0,Python,2015-01-04,"dask,numpy,pandas,pydata,python,scikit-learn,scipy",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；13,871 stars，最近可用 OpenRank 2.63，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
634081686,mlc-ai/mlc-llm,Universal LLM Deployment Engine with ML Compilation,23185,159,2137,345,Apache-2.0,false,2026-08-17,1.03,1.47,"[7.37,7.69,4.76,4.54,5.01,2.79,3.66,4.79,2.26,1.77,null,1.03]","[7.69, 4.76, 4.54, 5.01, 2.79, 3.66, 4.79, 2.26, 1.77, null, 1.03, 1.47]",2,Python,2023-04-29,"language-model,llm,machine-learning-compilation,tvm",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；23,003 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
660551251,FoundationAgents/MetaGPT,"🌟 The Multi-Agent Framework: First AI Software Company, Towards Natural Language Programming",70590,116,8962,139,MIT,false,2026-01-21,,1.19,"[3.09,2.25,0.99,3.63,2.51,2.53,2.07,1.85,1.17,0.84,0.96,null]","[2.25, 0.99, 3.63, 2.51, 2.53, 2.07, 1.85, 1.17, 0.84, 0.96, null, 1.19]",2,Python,2023-06-30,"agent,gpt,llm,metagpt,multi-agent",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；69,563 stars，最近可用 OpenRank 0.96，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
621799276,chatchat-space/Langchain-Chatchat,"Langchain-Chatchat（原Langchain-ChatGLM）基于 Langchain 与 ChatGLM, Qwen 与 Llama 等语言模型的 RAG 与 Agent 应用 | Langchain-Chatchat (formerly langchain-ChatGLM), local knowledge based LLM (like ChatGLM, Qwen and Llama) RAG and Agent app with langchain ",38665,155,6265,31,Apache-2.0,false,2025-11-10,,,"[5.44,4.95,2.18,3.87,2.76,2.16,1.44,null,0.62,0.46,null,null]","[4.95, 2.18, 3.87, 2.76, 2.16, 1.44, null, 0.62, 0.46, null, null, null]",0,Python,2023-03-31,"chatbot,chatchat,chatglm,chatgpt,embedding,faiss,fastchat,gpt,knowledge-base,langchain,langchain-chatglm,llama,llm,milvus,ollama,qwen,rag,retrieval-augmented-generation,streamlit,xinference",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；38,482 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
784181462,ItzCrazyKns/Vane,Vane is an AI-powered answering engine.,36904,48,4099,354,MIT,false,2026-09-01,,,"[10.81,11.36,11.36,7.94,7.58,8.92,8.13,7.62,4.35,2.42,null,null]","[11.36, 11.36, 7.94, 7.58, 8.92, 8.13, 7.62, 4.35, 2.42, null, null, null]",0,TypeScript,2024-04-09,"ai-agents,ai-search-engine,answering-engine,artificial-intelligence,llm,machine-learning,open-source-ai-search-engine,perplexica,rag,search-engine,searxng,searxng-copilot,self-hosted-ai,vane",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；35,896 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
156939672,microsoft/onnxruntime,"ONNX Runtime: cross-platform, high performance ML inferencing and training accelerator",21927,399,4250,1734,MIT,false,2026-09-24,46.91,38.45,"[202.54,182.08,165.3,132.78,133.97,133.76,111.79,104.25,86.21,81.37,63.96,46.91]","[182.08, 165.3, 132.78, 133.97, 133.76, 111.79, 104.25, 86.21, 81.37, 63.96, 46.91, 38.45]",22,C++,2018-11-10,"ai-framework,deep-learning,hardware-acceleration,machine-learning,neural-networks,onnx,pytorch,scikit-learn,tensorflow",add,Model Infra,Serving · Inference,跨平台推理 runtime 仍是 CPU、GPU 和边缘部署的重要基线，当前 21k+ stars、最近可用 OpenRank 64.52。,项目覆盖广泛 ML workload，不专属于 LLM；在图中只表达通用 inference runtime。,ok,,
627480054,eosphoros-ai/DB-GPT,open-source agentic AI data assistant for the next generation of AI + Data products.,20046,187,2944,457,MIT,false,2026-09-16,22.22,17.54,"[20.0,14.24,11.8,9.51,11.22,8.58,5.83,8.16,11.1,15.74,13.54,22.22]","[14.24, 11.8, 9.51, 11.22, 8.58, 5.83, 8.16, 11.1, 15.74, 13.54, 22.22, 17.54]",36,Python,2023-04-13,"agents,bgi,database,deepseek,gpt,gpt-4,hacktoberfest,llm,private,rag,security,vicuna",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；19,583 stars，最近可用 OpenRank 14.83，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
523007292,openxla/xla,"A machine learning compiler for GPUs, CPUs, and ML accelerators",4554,418,949,1319,Apache-2.0,false,2026-09-24,11.48,15.31,"[59.0,61.63,66.63,55.51,50.35,45.73,36.61,36.97,36.79,33.35,23.09,11.48]","[61.63, 66.63, 55.51, 50.35, 45.73, 36.61, 36.97, 36.79, 33.35, 23.09, 11.48, 15.31]",10,C++,2022-08-09,,add,Model Infra,Pre-Train · Compiler & accelerator,Apache-2.0 的跨 GPU、CPU 和专用加速器编译基础设施，补足当前编译器层只展示 kernel 与算子库的问题。,star 规模不大；保留依据是结构代表性和开放编译栈位置。,ok,,
703411624,letta-ai/letta,Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.,24867,139,2628,0,Apache-2.0,false,2026-09-10,1.89,0.77,"[22.51,27.1,18.43,9.93,7.95,6.91,3.89,3.58,3.3,2.1,1.04,1.89]","[27.1, 18.43, 9.93, 7.95, 6.91, 3.89, 3.58, 3.3, 2.1, 1.04, 1.89, 0.77]",5,,2023-10-11,"ai,ai-agents,llm,llm-agent",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；23,997 stars，最近可用 OpenRank 0.86，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
631254164,songquanpeng/one-api,"LLM API 管理 & 分发系统，支持 OpenAI、Azure、Anthropic Claude、Google Gemini、DeepSeek、字节豆包、ChatGLM、文心一言、讯飞星火、通义千问、360 智脑、腾讯混元等主流模型，统一 API 适配，可用于 key 管理与二次分发。单可执行文件，提供 Docker 镜像，一键部署，开箱即用。LLM API management & key redistribution system, unifying multiple providers under a single API. Single binary, Docker-ready, with an English UI.",37017,130,6868,1038,MIT,false,2026-01-09,,0.98,"[3.63,2.07,2.34,1.74,1.53,1.1,0.87,0.71,1.29,0.88,0.63,null]","[2.07, 2.34, 1.74, 1.53, 1.1, 0.87, 0.71, 1.29, 0.88, 0.63, null, 0.98]",1,JavaScript,2023-04-22,"api,api-gateway,azure-openai-api,chatgpt,claude,ernie-bot,gemini,gpt,openai,openai-api,proxy",remove,Model Infra,Model API gateways,与 New API、LiteLLM 高度重叠，最近可用 OpenRank 只有 0.64。,保留 New API 作为该产品谱系的当前代表。,ok,,
33015583,keras-team/keras,Deep Learning for humans,64336,407,19808,262,Apache-2.0,false,2026-09-24,14.86,14.66,"[47.95,46.14,48.76,44.02,41.4,41.66,40.32,42.76,35.03,29.29,20.69,14.86]","[46.14, 48.76, 44.02, 41.4, 41.66, 40.32, 42.76, 35.03, 29.29, 20.69, 14.86, 14.66]",11,Python,2015-03-28,"data-science,deep-learning,jax,machine-learning,neural-networks,python,pytorch,tensorflow",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；64,185 stars，最近可用 OpenRank 20.25，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
654122609,InternLM/lmdeploy,"LMDeploy is a toolkit for compressing, deploying, and serving LLMs.",8095,155,755,597,Apache-2.0,false,2026-09-23,,20.65,"[52.56,52.64,45.03,44.75,40.65,40.19,43.32,42.04,40.88,34.58,27.19,null]","[52.64, 45.03, 44.75, 40.65, 40.19, 43.32, 42.04, 40.88, 34.58, 27.19, null, 20.65]",1,Python,2023-06-15,"codellama,cuda-kernels,deepspeed,fastertransformer,internlm,llama,llama2,llama3,llm,llm-inference,turbomind",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；7,980 stars，最近可用 OpenRank 19.06，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
638629097,Aider-AI/aider,aider is AI pair programming in your terminal,49152,172,4987,1886,Apache-2.0,false,2026-05-22,2.42,4.56,"[45.5,29.05,22.27,17.59,15.89,11.37,8.77,7.19,8.23,5.43,4.75,2.42]","[29.05, 22.27, 17.59, 15.89, 11.37, 8.77, 7.19, 8.23, 5.43, 4.75, 2.42, 4.56]",3,Python,2023-05-09,"anthropic,chatgpt,claude-3,cli,command-line,gemini,gpt-3,gpt-35-turbo,gpt-4,gpt-4o,llama,openai,sonnet",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；47,754 stars，最近可用 OpenRank 4.17，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
359952601,pgvector/pgvector,Open-source vector similarity search for Postgres,23148,23,1329,17,NOASSERTION,false,2026-09-22,,,"[5.11,5.76,3.43,2.95,4.88,5.08,2.85,3.48,1.52,0.64,3.77,null]","[5.76, 3.43, 2.95, 4.88, 5.08, 2.85, 3.48, 1.52, 0.64, 3.77, null, null]",0,C,2021-04-20,"approximate-nearest-neighbor-search,nearest-neighbor-search",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；22,373 stars，最近可用 OpenRank 3.77，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
151636194,triton-inference-server/server,The Triton Inference Server provides an optimized cloud and edge inferencing solution. ,11008,139,1839,894,BSD-3-Clause,false,2026-09-23,3.18,8.99,"[20.19,20.44,18.15,15.02,15.62,11.43,11.31,11.74,11.85,10.73,5.54,3.18]","[20.44, 18.15, 15.02, 15.62, 11.43, 11.31, 11.74, 11.85, 10.73, 5.54, 3.18, 8.99]",2,Python,2018-10-04,"cloud,datacenter,deep-learning,edge,gpu,inference,machine-learning",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；10,875 stars，最近可用 OpenRank 6.21，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
612344730,ChatGPTNextWeb/NextChat,"✨ Zero-config AI chat assistant. No API key needed — sign up and instantly chat with GPT-5, Claude 4, Gemini 2.5, DeepSeek & 100+ top models. Pay-as-you-go saves you more. Available on Web, iOS, macOS, Android, Linux, Windows.",88809,262,58989,866,MIT,false,2026-08-11,,0.74,"[10.94,8.37,3.71,4.41,3.97,2.53,2.84,1.34,1.12,0.58,null,null]","[8.37, 3.71, 4.41, 3.97, 2.53, 2.84, 1.34, 1.12, 0.58, null, null, 0.74]",1,TypeScript,2023-03-10,"calclaude,chatgpt,claude,cross-platform,desktop,fe,gemini,gemini-pro,gemini-server,gemini-ultra,gpt-4o,groq,nextjs,ollama,react,tauri,tauri-app,vercel,webui",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；88,563 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
778431525,microsoft/graphrag,A modular graph-based Retrieval-Augmented Generation (RAG) system,36092,51,3800,38,MIT,false,2026-09-24,,1.23,"[17.96,16.86,15.53,11.52,9.94,8.71,10.85,5.47,3.14,2.55,null,null]","[16.86, 15.53, 11.52, 9.94, 8.71, 10.85, 5.47, 3.14, 2.55, null, null, 1.23]",7,Python,2024-03-27,"gpt,gpt-4,gpt4,graphrag,llm,llms,rag",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；34,955 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
547806116,huggingface/text-generation-inference,Large Language Model Text Generation Inference,10884,145,1292,324,Apache-2.0,true,2026-03-21,,,"[8.75,5.29,2.78,1.07,0.63,0.91,0.44,0.45,null,null,null,null]","[5.29, 2.78, 1.07, 0.63, 0.91, 0.44, 0.45, null, null, null, null, null]",0,Python,2022-10-08,"bloom,deep-learning,falcon,gpt,inference,nlp,pytorch,starcoder,transformer",omit,,,仓库当前已归档，不进入下一版 landscape。,历史影响力仍可在报告正文中保留。,ok,,
654124617,open-compass/opencompass,"OpenCompass is an LLM evaluation platform, supporting a wide range of models from OpenAI, Anthropic, Gemini, Qwen, GLM, DeepSeek, etc, across 100+ datasets covering knowledge, reasoning, coding, science, language, long-context, and safety.",7472,183,868,405,Apache-2.0,false,2026-09-23,,3.14,"[16.46,16.12,9.25,8.91,7.24,5.33,5.22,4.42,4.08,null,1.41,null]","[16.12, 9.25, 8.91, 7.24, 5.33, 5.22, 4.42, 4.08, null, 1.41, null, 3.14]",1,Python,2023-06-15,"benchmark,evaluation,large-language-model,llm,llm-eval",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；7,241 stars，最近可用 OpenRank 1.73，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
607441698,lancedb/lancedb,Developer-friendly OSS embedded retrieval library for multimodal AI. Search More; Manage Less.,11519,247,1063,683,Apache-2.0,false,2026-09-24,10.88,15.99,"[18.3,19.24,16.35,17.11,15.41,13.93,16.5,18.28,17.37,12.79,13.63,10.88]","[19.24, 16.35, 17.11, 15.41, 13.93, 16.5, 18.28, 17.37, 12.79, 13.63, 10.88, 15.99]",15,Rust,2023-02-28,"approximate-nearest-neighbor-search,image-search,nearest-neighbor-search,recommender-system,search-engine,semantic-search,similarity-search,vector-database",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；11,014 stars，最近可用 OpenRank 14.40，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
619959033,nomic-ai/gpt4all,GPT4All: Run Local LLMs on Any Device. Open-source and available for commercial use.,77386,115,8288,773,MIT,false,2025-05-27,,0.64,"[3.16,2.9,2.09,1.13,1.03,0.96,3.18,1.57,null,null,null,null]","[2.9, 2.09, 1.13, 1.03, 0.96, 3.18, 1.57, null, null, null, null, 0.64]",0,C++,2023-03-27,"ai-chat,llm-inference",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；77,397 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
615882673,lm-sys/FastChat,"An open platform for training, serving, and evaluating large language models. Release repo for Vicuna and Chatbot Arena.",39551,240,4776,1042,Apache-2.0,false,2026-05-01,,,"[3.11,2.48,2.66,2.52,2.97,2.57,3.38,4.58,7.01,6.11,null,null]","[2.48, 2.66, 2.52, 2.97, 2.57, 3.38, 4.58, 7.01, 6.11, null, null, null]",1,Python,2023-03-19,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；39,504 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
45717250,tensorflow/tensorflow,An Open Source Machine Learning Framework for Everyone,200275,411,77176,3341,Apache-2.0,false,2026-09-24,9.56,11.51,"[66.68,58.23,41.32,47.75,40.07,45.63,37.02,35.42,26.45,20.34,14.28,9.56]","[58.23, 41.32, 47.75, 40.07, 45.63, 37.02, 35.42, 26.45, 20.34, 14.28, 9.56, 11.51]",26,C++,2015-11-07,"deep-learning,deep-neural-networks,distributed,machine-learning,ml,neural-network,python,tensorflow",add,Model Infra,Pre-Train · Framework & parallel,通用训练框架的历史基线仍应保留，尤其在 NeMo 仓库改名为 Speech 后，原位置已经不再准确。,近期 OpenRank 明显低于 PyTorch、JAX；入图表示生态基线，不解释成增长项目。,ok,,
942114501,camel-ai/owl,🦉 OWL: Optimized Workforce Learning for General Multi-Agent Assistance in Real-World Task Automation,20146,39,2299,119,,false,2026-09-20,,0.42,"[3.27,null,null,1.12,0.66,1.29,1.01,0.51,0.73,0.6,null,null]","[null, null, 1.12, 0.66, 1.29, 1.01, 0.51, 0.73, 0.6, null, null, 0.42]",0,Python,2025-03-03,"agent,artificial-intelligence,multi-agent-systems,task-automation,web-interaction",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；20,070 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
862449441,stackblitz/bolt.new,"Prompt, run, edit, and deploy full-stack web applications. -- bolt.new -- Help Center: https://support.bolt.new/ -- Community Support: https://discord.com/invite/stackblitz",16562,16,14800,8188,MIT,false,2024-12-17,,0.46,"[11.56,3.82,1.53,0.53,0.47,0.37,null,0.56,null,null,null,null]","[3.82, 1.53, 0.53, 0.47, 0.37, null, 0.56, null, null, null, null, 0.46]",0,TypeScript,2024-09-24,"ai-playground,anthropic,chatbot,webcontainers-api",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；16,480 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
662766482,The-PR-Agent/pr-agent,🚀 PR Agent: The Original Open-Source PR Reviewer. This project is not the Qodo free tier.,13137,347,1885,62,MIT,false,2026-09-24,5.22,9.99,"[26.33,20.65,14.14,12.1,4.42,8.76,10.23,6.85,5.88,7.5,6.36,5.22]","[20.65, 14.14, 12.1, 4.42, 8.76, 10.23, 6.85, 5.88, 7.5, 6.36, 5.22, 9.99]",3,Python,2023-07-05,"ai,anthropic,code-review,coding-assistant,devtools,ollama,openai,pull-request,pull-requests",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；12,276 stars，最近可用 OpenRank 6.24，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
943934999,FoundationAgents/OpenManus,"No fortress, purely open ground.  OpenManus is Coming.",58401,54,10134,452,MIT,false,2026-08-22,,,"[6.24,6.57,8.27,6.89,5.0,6.67,2.15,1.32,1.74,1.09,0.41,null]","[6.57, 8.27, 6.89, 5.0, 6.67, 2.15, 1.32, 1.74, 1.09, 0.41, null, null]",1,Python,2025-03-06,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；57,758 stars，最近可用 OpenRank 0.35，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
614764248,TabbyML/tabby,Self-hosted AI coding assistant,33892,117,1786,339,NOASSERTION,false,2026-06-30,,0.54,"[8.75,4.82,2.8,2.93,2.62,1.6,2.82,1.34,0.71,0.55,null,null]","[4.82, 2.8, 2.93, 2.62, 1.6, 2.82, 1.34, 0.71, 0.55, null, null, 0.54]",0,Rust,2023-03-16,"ai,codegen,coding-assistant,coding-language,developer-experience,developer-tools,gen-ai,ide,llms",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；33,790 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1024554267,NousResearch/hermes-agent,The agent that grows with you,248624,398,52595,43707,MIT,false,2026-09-24,350.21,277.91,"[null,null,null,null,null,null,14.46,121.65,501.07,609.35,518.86,350.21]","[null, null, null, null, null, 14.46, 121.65, 501.07, 609.35, 518.86, 350.21, 277.91]",549,Python,2025-07-22,"ai,ai-agent,ai-agents,anthropic,chatgpt,claude,claude-code,codex,hermes,hermes-agent,llm,nous-research,openai",keep,Agent Infra,Personal AI assistants,"当前图中的 Personal AI assistants 代表项目；仓库仍在维护，221,762 stars，最近可用 OpenRank 503.71。",产品边界普遍较宽，只在这一分区出现，避免与 chatbot 和 framework 重复。,ok,,
1073224795,obra/superpowers,An agentic skills framework & software development methodology that works.,291057,37,26038,400,MIT,false,2026-09-22,5.12,12.62,"[null,null,10.5,11.82,16.67,32.59,30.14,38.04,32.21,21.43,12.9,5.12]","[null, 10.5, 11.82, 16.67, 32.59, 30.14, 38.04, 32.21, 21.43, 12.9, 5.12, 12.62]",5,Shell,2025-10-09,"ai,brainstorming,coding,obra,sdlc,skills,subagent-driven-development,superpowers",keep,Agent Infra,Coding workflows & harnesses,"当前图中的 Coding harnesses 代表项目；仓库仍在维护，262,543 stars，最近可用 OpenRank 12.59。",harness 热度容易受短期传播影响，需要持续检查外部贡献和真实复用。,ok,,
1213049020,browser-use/browser-harness,Browser Harness | Self-healing harness that enables LLMs to complete any task.,18086,69,1764,394,MIT,false,2026-09-12,,5.7,"[null,null,null,null,null,null,null,null,10.2,7.31,2.81,null]","[null, null, null, null, null, null, null, 10.2, 7.31, 2.81, null, 5.7]",1,Python,2026-04-17,"ai-agent,browser-agent,browser-automation,browser-use,browser-use-box,browser-use-cloud,cdp,cloud-browser,llm,persistent-browser,playwright,telegram-agent,vps-agent,web-automation",remove,Agent Infra,"Tools, web & computer use",与同组织 browser-use 主项目重叠，Tool & browser use 区只保留一个主 logo。,其 self-healing harness 能力可以作为 browser-use 的项目演化注释。,ok,,
1139971460,rtk-ai/rtk,"CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies",81639,149,5166,1575,Apache-2.0,false,2026-09-24,12.05,16.08,"[null,null,null,null,null,1.36,16.06,40.04,40.12,34.53,21.48,12.05]","[null, null, null, null, 1.36, 16.06, 40.04, 40.12, 34.53, 21.48, 12.05, 16.08]",20,Rust,2026-01-22,"agentic-coding,ai-coding,anthropic,claude-code,cli,command-line-tool,cost-reduction,developer-tools,llm,open-source,productivity,rust,token-optimization",remove,Model Infra,Model API gateways,项目是开发命令 token 压缩 CLI，并非 Model API gateway；现有位置属于误分类。,它可以继续作为 coding workflow 工具观察，但暂不占主图位置。,ok,,
1141782198,jamiepine/voicebox,"The open-source AI voice studio. Clone, dictate, create.",55580,69,6928,703,MIT,false,2026-08-09,1.83,3.35,"[null,null,null,null,null,4.66,16.81,15.15,17.14,7.87,3.56,1.83]","[null, null, null, null, 4.66, 16.81, 15.15, 17.14, 7.87, 3.56, 1.83, 3.35]",5,TypeScript,2026-01-25,"ai,cuda,mlx,qwen3-tts,qwen3-tts-ui,voice-ai,voice-clone,whisper",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；47,182 stars，最近可用 OpenRank 3.35，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1174820787,karpathy/autoresearch,AI agents running research on single-GPU nanochat training automatically,96716,9,13524,194,,false,2026-03-26,,0.67,"[null,null,null,null,null,null,null,18.34,7.03,2.89,2.3,null]","[null, null, null, null, null, null, 18.34, 7.03, 2.89, 2.3, null, 0.67]",0,Python,2026-03-06,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；92,224 stars，最近可用 OpenRank 2.28，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1148052330,Yeachan-Heo/oh-my-codex,"OmX - Oh My codeX: Your codex is not alone. Add hooks, agent teams, HUDs, and so much more.",33352,92,2543,1,MIT,false,2026-09-23,12.36,11.31,"[null,null,null,null,null,null,8.66,17.82,32.44,29.22,14.18,12.36]","[null, null, null, null, null, 8.66, 17.82, 32.44, 29.22, 14.18, 12.36, 11.31]",9,TypeScript,2026-02-02,,remove,Agent Infra,Coding workflows & harnesses,与 ECC、Superpowers、oh-my-openagent 同属 coding harness 增强层，当前版面无需四个相似配置栈。,保留在项目池中观察其独立社区和跨 agent 支持。,ok,,
1201656210,MemPalace/mempalace,The best-benchmarked open-source AI memory system. And it's free.,59260,143,7568,752,MIT,false,2026-09-23,7.44,11.88,"[null,null,null,null,null,null,null,null,57.18,34.66,18.72,7.44]","[null, null, null, null, null, null, null, 57.18, 34.66, 18.72, 7.44, 11.88]",5,Python,2026-04-05,"ai,chromadb,llm,mcp,memory,python",remove,Agent Infra,"Memory, knowledge & context",短期关注度很高，但与 Mem0、Hindsight 的长期记忆定位重叠，独立采用证据仍需观察。,保留为观察项目；若持续贡献和外部采用站稳，可替换当前 memory 代表。,ok,,
1035029907,earendil-works/pi,"AI agent toolkit: unified LLM API, agent loop, TUI, coding agent CLI",109127,287,13871,215,MIT,false,2026-09-24,43.89,35.31,"[null,null,null,7.22,30.51,67.46,72.04,107.01,119.27,111.81,72.68,43.89]","[null, null, 7.22, 30.51, 67.46, 72.04, 107.01, 119.27, 111.81, 72.68, 43.89, 35.31]",18,TypeScript,2025-08-09,,keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，79,543 stars，最近可用 OpenRank 69.59。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
1170821064,paperclipai/paperclip,The open-source app everyone uses to manage agents at work,81743,202,14986,5618,MIT,false,2026-09-24,49.68,45.36,"[null,null,null,null,null,null,null,87.47,122.24,104.72,72.16,49.68]","[null, null, null, null, null, null, 87.47, 122.24, 104.72, 72.16, 49.68, 45.36]",98,TypeScript,2026-03-02,,keep,Agent Infra,Multi-agent orchestration,"当前图中的 Multi-agent orchestration 代表项目；仓库仍在维护，74,970 stars，最近可用 OpenRank 67.51。",多 agent 概念变化快，保留项目需要持续显示独立编排能力。,ok,,
798235565,ant-design/x,Craft AI-driven interface effortlessly🤖,4799,90,1161,173,,false,2026-09-20,15.82,12.75,"[26.98,26.07,22.84,25.85,41.89,40.1,32.86,24.33,22.21,17.07,15.79,15.82]","[26.07, 22.84, 25.85, 41.89, 40.1, 32.86, 24.33, 22.21, 17.07, 15.79, 15.82, 12.75]",37,TypeScript,2024-05-09,"ai,antd,chat-ui,copilot,react",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；4,695 stars，最近可用 OpenRank 18.81，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
946498681,neovateai/neovate-code,"Neovate Code is a code agent to enhance your development. You can use it to generate code, fix bugs, review code, add tests, and more. You can run it in interactive mode or headless mode.",1558,34,154,56,MIT,false,2026-03-24,0.95,0.72,"[12.16,19.38,22.1,26.73,25.5,17.45,13.47,7.97,null,2.98,2.0,0.95]","[19.38, 22.1, 26.73, 25.5, 17.45, 13.47, 7.97, null, 2.98, 2.0, 0.95, 0.72]",1,TypeScript,2025-03-11,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；1,554 stars，最近可用 OpenRank 2.34，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
655515393,CopilotKit/CopilotKit,"The Frontend Stack for Agents & Generative UI. React, Angular, Mobile, Slack, and more.  Makers of the AG-UI Protocol",37528,219,4657,289,MIT,false,2026-09-24,31.48,28.95,"[43.2,39.96,35.11,27.09,35.77,43.57,41.14,35.37,37.0,31.01,31.19,31.48]","[39.96, 35.11, 27.09, 35.77, 43.57, 41.14, 35.37, 37.0, 31.01, 31.19, 31.48, 28.95]",24,TypeScript,2023-06-19,"agent,agent-native,agentic-ai,agents,ai,ai-agent,ai-assistant,assistant,assistant-chat-bots,copilot,copilot-chat,generative-ui,js,llm,nextjs,open-source,react,reactjs,ts,typescript",keep,Agent Infra,Code-first frameworks,"当前图中的 Code-first frameworks 代表项目；仓库仍在维护，36,339 stars，最近可用 OpenRank 35.83。",该区按语言和 runtime 多样性取样，不追求把所有 framework 都放进来。,ok,,
470472185,koordinator-sh/koordinator,"A QoS-based scheduling system brings optimal layout and status to workloads such as microservices, web services, big data jobs, AI jobs, etc.",1759,128,481,231,Apache-2.0,false,2026-09-23,4.18,6.9,"[20.15,19.39,11.23,12.29,14.27,13.26,8.94,8.49,5.54,8.07,7.45,4.18]","[19.39, 11.23, 12.29, 14.27, 13.26, 8.94, 8.49, 5.54, 8.07, 7.45, 4.18, 6.9]",4,Go,2022-03-16,"cloud-native,colocation,cost-optimization,kubernetes,scheduling,workload-consolidation",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；1,728 stars，最近可用 OpenRank 8.48，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
763774926,EricLBuehler/mistral.rs,"Fast, flexible LLM inference",7718,95,714,394,MIT,false,2026-09-24,,6.01,"[20.19,14.74,11.14,11.39,11.78,12.26,15.15,11.56,10.3,7.85,7.35,null]","[14.74, 11.14, 11.39, 11.78, 12.26, 15.15, 11.56, 10.3, 7.85, 7.35, null, 6.01]",1,Rust,2024-02-26,"llm,rust,uqff",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；7,537 stars，最近可用 OpenRank 8.30，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1080442728,oceanbase/seekdb,"The AI-Native Search Database. Best for agent storage, it unifies vector, text, structured, and semi-structured data into a single engine. This all-in-one database makes agents smarter, easier to run, and more stable.",3056,35,358,373,Apache-2.0,false,2026-09-24,6.29,10.67,"[null,null,null,5.13,6.88,6.59,7.29,7.07,5.55,6.32,6.44,6.29]","[null, null, 5.13, 6.88, 6.59, 7.29, 7.07, 5.55, 6.32, 6.44, 6.29, 10.67]",3,C++,2025-10-21,"ai-agents,copy-on-write,embedded-database,full-text-search,hnsw,hybrid-search,langchain,llamaindex,mysql,oceanbase,python,rag,sql,vector-databases",keep,Agent Infra,"Memory, knowledge & context","当前图中的 Memory, knowledge & context 代表项目；仓库仍在维护，2,851 stars，最近可用 OpenRank 6.25。",数据库、RAG 和 memory 边界容易混淆，图中要标清每个项目承担的层次。,ok,,
1032459340,Fission-AI/OpenSpec,Spec-driven development (SDD) for AI coding assistants.,70176,120,4801,222,MIT,false,2026-09-23,4.67,4.55,"[0.63,2.86,13.09,13.88,16.82,17.18,16.17,10.4,12.16,8.77,6.96,4.67]","[2.86, 13.09, 13.88, 16.82, 17.18, 16.17, 10.4, 12.16, 8.77, 6.96, 4.67, 4.55]",4,TypeScript,2025-08-05,"ai,context-engineering,engineering,planning,prd,sdd,sdlc,spec,spec-driven-development,specification",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；62,922 stars，最近可用 OpenRank 8.10，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1028070615,NevaMind-AI/memU,Personal memory across agents,14427,44,1070,122,NOASSERTION,false,2026-09-21,2.33,3.49,"[4.18,4.72,4.49,4.76,5.94,14.32,13.24,6.56,3.27,1.0,2.46,2.33]","[4.72, 4.49, 4.76, 5.94, 14.32, 13.24, 6.56, 3.27, 1.0, 2.46, 2.33, 3.49]",2,Python,2025-07-29,"agent-memory,claude-skills,harness,loop-engineering,mcp,memory,openclaw,openclaw-skills,sandbox,skills",remove,Agent Infra,"Memory, knowledge & context",个人 memory 定位与 Mem0、Hindsight 重叠，最近可用 OpenRank 只有 2.40。,不因短期 stars 单独增加 logo。,ok,,
742244656,agentscope-ai/agentscope,"Build and run agents you can see, understand and trust.",32300,141,3555,435,Apache-2.0,false,2026-09-24,6.71,10.98,"[10.86,25.58,21.5,23.74,19.82,21.35,19.63,15.53,11.86,9.95,10.12,6.71]","[25.58, 21.5, 23.74, 19.82, 21.35, 19.63, 15.53, 11.86, 9.95, 10.12, 6.71, 10.98]",4,Python,2024-01-12,"agent,chatbot,large-language-models,llm,llm-agent,mcp,multi-agent,multi-modal,react-agent,realtime-agent,voice-agent,voice-assistant",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；28,351 stars，最近可用 OpenRank 11.34，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
384219990,warpdotdev/warp,"Warp is an agentic development environment, born out of the terminal.",65157,166,5584,5289,AGPL-3.0,false,2026-09-24,53.23,40.12,"[117.66,141.6,124.86,98.01,73.53,63.26,58.45,43.53,59.96,100.54,81.1,53.23]","[141.6, 124.86, 98.01, 73.53, 63.26, 58.45, 43.53, 59.96, 100.54, 81.1, 53.23, 40.12]",37,Rust,2021-07-08,"bash,linux,macos,rust,shell,terminal,wasm,zsh",keep,Agent Infra,Development sandboxes,"把终端、Agent 执行与开发环境整合在一起，更适合作为 Agent Runtime Infra 中的执行环境代表。",产品入口仍与 Agentic coding 有交叉；本图将它放在 Runtime，强调任务执行环境而非独立 coding harness。,ok,,
1147094660,HKUDS/nanobot,"Ultra-lightweight, open-source, self-hosted personal AI agent framework in Python with WebUI, tools, memory, MCP, multi-agent workflows, automation, and chat apps",48545,384,8576,790,MIT,false,2026-09-24,36.96,29.2,"[null,null,null,null,null,null,98.88,114.32,98.65,69.55,46.49,36.96]","[null, null, null, null, null, 98.88, 114.32, 98.65, 69.55, 46.49, 36.96, 29.2]",16,Python,2026-02-01,"agent-framework,ai-agent,ai-agents,chatbot,chatops,discord-bot,llm-agents,llms,local-llm,mcp,model-context-protocol,multi-agent,openai-compatible,openclaw,personal-ai-assistant,python,self-hosted,telegram-bot-ai-assistant,webui,workflow-automation",keep,Agent Infra,Personal AI assistants,"当前图中的 Personal AI assistants 代表项目；仓库仍在维护，46,331 stars，最近可用 OpenRank 44.73。",产品边界普遍较宽，只在这一分区出现，避免与 chatbot 和 framework 重复。,ok,,
1130809465,Yeachan-Heo/oh-my-claudecode,Teams-first Multi-agent orchestration for Claude Code,39338,144,3514,2,MIT,false,2026-09-24,4.05,9.21,"[null,null,null,null,null,13.77,30.26,30.85,32.39,16.32,8.01,4.05]","[null, null, null, null, 13.77, 30.26, 30.85, 32.39, 16.32, 8.01, 4.05, 9.21]",4,TypeScript,2026-01-09,"agentic-coding,ai-agents,automation,claude,claude-code,multi-agent-systems,oh-my-opencode,opencode,parallel-execution,vibe-coding",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；38,129 stars，最近可用 OpenRank 7.16，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1133707766,multica-ai/multica,Make humans and AI agents work as one team — open-source and self-hostable.,51293,320,6646,1671,NOASSERTION,false,2026-09-24,30.63,34.58,"[null,null,null,null,null,1.92,1.77,3.84,44.33,54.88,43.33,30.63]","[null, null, null, null, 1.92, 1.77, 3.84, 44.33, 54.88, 43.33, 30.63, 34.58]",33,Go,2026-01-13,,keep,Agent Infra,Multi-agent orchestration,"当前图中的 Multi-agent orchestration 代表项目；仓库仍在维护，42,351 stars，最近可用 OpenRank 38.68。",多 agent 概念变化快，保留项目需要持续显示独立编排能力。,ok,,
979115477,bytedance/deer-flow,"An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.",82941,452,11479,887,MIT,false,2026-09-24,218.2,211.17,"[40.56,37.96,31.46,28.94,27.61,25.67,32.37,136.15,203.53,190.8,180.87,218.2]","[37.96, 31.46, 28.94, 27.61, 25.67, 32.37, 136.15, 203.53, 190.8, 180.87, 218.2, 211.17]",181,Python,2025-05-07,"agent,agentic,agentic-framework,agentic-workflow,ai,ai-agents,deep-research,harness,langchain,langgraph,langmanus,llm,multi-agent,nodejs,podcast,python,superagent,typescript",keep,Agent Infra,Multi-agent orchestration,"当前图中的 Multi-agent orchestration 代表项目；仓库仍在维护，78,057 stars，最近可用 OpenRank 172.36。",多 agent 概念变化快，保留项目需要持续显示独立编排能力。,ok,,
979886407,IBM/mcp-context-forge,"An AI Gateway, registry, and proxy that sits in front of any MCP, A2A, or REST/gRPC APIs, exposing a unified endpoint with centralized discovery, guardrails and management. Optimizes Agent & Tool calling, and supports plugins.",4523,167,887,972,Apache-2.0,false,2026-09-24,30.35,23.36,"[27.8,36.63,33.62,36.23,39.35,46.45,57.66,60.33,55.23,47.07,32.0,30.35]","[36.63, 33.62, 36.23, 39.35, 46.45, 57.66, 60.33, 55.23, 47.07, 32.0, 30.35, 23.36]",13,Python,2025-05-08,"agents,ai,api-gateway,asyncio,authentication-middleware,devops,docker,fastapi,federation,gateway,generative-ai,jwt,kubernetes,llm-agents,mcp,model-context-protocol,observability,prompt-engineering,python,tools",add,Agent Infra,Protocols & interoperability,MCP gateway、registry、治理与可观测属于 agent 协议控制面，不是模型 API 聚合。,与 AgentGateway 有交叉；保留两个是为了区分流量代理和 MCP 管理平台，不再继续增加同类 gateway。,ok,,
1132001614,vercel-labs/agent-browser,Browser automation CLI for AI agents,43150,115,2894,815,Apache-2.0,false,2026-09-23,6.29,7.74,"[null,null,null,null,null,29.12,24.14,30.93,19.68,8.25,5.76,6.29]","[null, null, null, null, 29.12, 24.14, 30.93, 19.68, 8.25, 5.76, 6.29, 7.74]",7,Rust,2026-01-11,,keep,Agent Infra,"Tools, web & computer use","当前图中的 Tool & browser use 代表项目；仓库仍在维护，39,397 stars，最近可用 OpenRank 3.54。",browser/computer-use 项目迭代很快，需要按可复现能力和维护状态复核。,ok,,
1086419061,vectorize-io/hindsight,Hindsight: Agent Memory That Learns,27209,244,2569,168,MIT,false,2026-09-24,15.15,23.45,"[null,null,null,null,8.06,12.68,14.93,16.52,21.37,23.37,20.93,15.15]","[null, null, null, 8.06, 12.68, 14.93, 16.52, 21.37, 23.37, 20.93, 15.15, 23.45]",6,Python,2025-10-30,"agentic-ai,agents,ai-memory,memory",keep,Agent Infra,"Memory, knowledge & context","当前图中的 Memory, knowledge & context 代表项目；仓库仍在维护，18,857 stars，最近可用 OpenRank 22.31。",数据库、RAG 和 memory 边界容易混淆，图中要标清每个项目承担的层次。,ok,,
1136590548,affaan-m/ECC,"The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.",266587,356,39838,226,MIT,false,2026-09-22,14.71,21.08,"[null,null,null,null,null,9.36,17.63,28.26,30.4,25.38,20.77,14.71]","[null, null, null, null, 9.36, 17.63, 28.26, 30.4, 25.38, 20.77, 14.71, 21.08]",8,JavaScript,2026-01-18,"ai-agents,anthropic,claude,claude-code,developer-tools,llm,mcp,productivity",keep,Agent Infra,Coding workflows & harnesses,"当前图中的 Coding harnesses 代表项目；仓库仍在维护，234,546 stars，最近可用 OpenRank 18.20。",harness 热度容易受短期传播影响，需要持续检查外部贡献和真实复用。,ok,,
1165324947,agentscope-ai/QwenPaw,"Your Personal AI Assistant; easy to install, deploy on your own machine or on the cloud; supports multiple chat apps with easily extensible capabilities.",35261,239,3128,954,Apache-2.0,false,2026-09-24,35.86,39.14,"[null,null,null,null,null,null,14.74,119.18,107.06,76.59,60.29,35.86]","[null, null, null, null, null, 14.74, 119.18, 107.06, 76.59, 60.29, 35.86, 39.14]",20,TypeScript,2026-02-24,"agent,agent-harness,agentscope,ai-agent,ai-agents,chatbot,harness-engineering,llm-tools,llms,loop-engineering,mcp,personal-ai-assistant,self-hosted,skills,super-agent,webui",keep,Agent Infra,Personal AI assistants,"当前图中的 Personal AI assistants 代表项目；仓库仍在维护，29,630 stars，最近可用 OpenRank 57.61。",产品边界普遍较宽，只在这一分区出现，避免与 chatbot 和 framework 重复。,ok,,
1197990888,Gitlawb/openclaude,runs anywhere. uses anything,33515,150,9100,63,NOASSERTION,false,2026-09-22,11.77,13.5,"[null,null,null,null,null,null,null,null,45.26,35.99,19.5,11.77]","[null, null, null, null, null, null, null, 45.26, 35.99, 19.5, 11.77, 13.5]",8,TypeScript,2026-04-01,"ai,ai-agent,ai-tools,cli,coding",remove,Agent Infra,Agentic coding,项目定位和描述过于模糊，难以在已经拥挤的 coding agent 区解释其独立代表性。,若后续形成稳定产品边界和独立社区，再重新评估。,ok,,
1062458369,alibaba/page-agent,JavaScript in-page GUI agent. Control web interfaces with natural language.,29195,33,2622,97,MIT,false,2026-09-21,1.55,2.16,"[null,0.89,2.08,2.1,3.29,4.38,3.8,10.62,9.61,3.99,null,1.55]","[0.89, 2.08, 2.1, 3.29, 4.38, 3.8, 10.62, 9.61, 3.99, null, 1.55, 2.16]",1,TypeScript,2025-09-23,"agent,ai,ai-agents,browser-automation,javascript,mcp,typescript,web",keep,Agent Infra,"Tools, web & computer use","当前图中的 Tool & browser use 代表项目；仓库仍在维护，28,053 stars，最近窗口暂未产出 OpenRank。",browser/computer-use 项目迭代很快，需要按可复现能力和维护状态复核。,ok,,
1008713177,QwenLM/qwen-code,An open-source AI coding agent that lives in your terminal.,28109,420,3102,1480,Apache-2.0,false,2026-09-24,53.69,47.73,"[73.88,73.22,56.87,45.22,49.09,48.48,48.56,51.85,60.28,59.16,55.69,53.69]","[73.22, 56.87, 45.22, 49.09, 48.48, 48.56, 51.85, 60.28, 59.16, 55.69, 53.69, 47.73]",38,TypeScript,2025-06-26,"agentic,ai,ai-agent,ai-coding,cli,coding-agent,developer-tools,llm,mcp,qwen,terminal,typescript",keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，26,393 stars，最近可用 OpenRank 60.25。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
1028492186,eigent-ai/eigent,Eigent: The Open Source Cowork Desktop - Local and Free Alternative to Claude Cowork and Codex,15418,73,1835,235,Apache-2.0,false,2026-09-24,5.74,6.4,"[26.37,32.14,30.41,32.29,16.61,35.06,37.03,31.11,18.62,9.17,null,5.74]","[32.14, 30.41, 32.29, 16.61, 35.06, 37.03, 31.11, 18.62, 9.17, null, 5.74, 6.4]",12,TypeScript,2025-07-29,"agent-framework,agent-skills,agentic-ai,agentic-workflow,claude-cowork,claude-cowork-alternative,claude-cowork-free,desktop-agent,multi-agent-systems",omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；14,678 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
1166139870,RightNow-AI/openfang,Open-source Agent Operating System,18211,64,2289,124,Apache-2.0,false,2026-07-02,,1.99,"[null,null,null,null,null,null,10.45,53.57,31.05,14.46,6.69,null]","[null, null, null, null, null, 10.45, 53.57, 31.05, 14.46, 6.69, null, 1.99]",0,Rust,2026-02-24,"agent-framework,ai-agents,llm,mcp,open-source,openclaw,operating-system,rust",remove,Agent Infra,Code-first frameworks,Agent OS 定位较宽，与现有 code-first framework 和 personal assistant 项目重叠，最近可用 OpenRank 也偏低。,若形成清晰 runtime 边界，可在后续版本重新考虑。,ok,,
1153302402,openai/parameter-golf,Train the smallest LM you can that fits in 16MB. Best model wins!,5191,45,3262,1425,MIT,false,2026-05-04,,,"[null,null,null,null,null,null,null,61.24,71.92,38.32,null,null]","[null, null, null, null, null, null, 61.24, 71.92, 38.32, null, null, null]",3,Python,2026-02-09,,omit,,,"候选项目已复核，但在当前版面预算下暂不入图：未进入本版主图；5,176 stars，最近窗口暂未产出 OpenRank，但与已选项目功能重叠，或没有形成新的结构位置。",继续保留在项目池，后续按持续协作、独立采用和分类缺口复核。,ok,,
585860664,github/copilot-cli,GitHub Copilot CLI brings the power of Copilot coding agent directly to your terminal. ,11200,26,1937,2267,NOASSERTION,false,2026-09-23,9.93,17.47,"[null,31.68,58.09,53.02,42.3,63.74,71.58,69.06,59.67,48.84,23.08,9.93]","[31.68, 58.09, 53.02, 42.3, 63.74, 71.58, 69.06, 59.67, 48.84, 23.08, 9.93, 17.47]",7,Shell,2023-01-06,,keep,Agent Infra,Agentic coding,"当前图中的 Agentic coding 代表项目；仓库仍在维护，11,029 stars，最近可用 OpenRank 21.01。",该区变化最快，logo 数量控制在 12 个，后续按持续协作和产品独立性替换。,ok,,
1118085970,opensandbox-group/OpenSandbox,"Secure, Fast, and Extensible Sandbox runtime for AI agents.",15500,124,1424,132,Apache-2.0,false,2026-09-24,11.3,,"[null,null,null,null,2.51,6.15,11.54,18.27,14.99,14.59,14.2,11.3]","[null, null, null, 2.51, 6.15, 11.54, 18.27, 14.99, 14.59, 14.2, 11.3, null]",10,Python,2025-12-17,"ai,ai-agent,ai-infra,kubernetes,sandbox",keep,Agent Infra,Development sandboxes,"当前图中的 Development sandboxes 代表项目；仓库仍在维护，12,219 stars，最近可用 OpenRank 16.08。",编排层和隔离 runtime 不是一回事，主图说明中要保留这一区别。,ok,,
1037741738,RLinf/RLinf,RLinf: Reinforcement Learning Infrastructure for Embodied and Agentic AI,5365,123,749,270,Apache-2.0,false,2026-09-24,10.12,22.1,"[2.91,15.41,22.77,29.92,38.67,42.17,36.58,44.96,42.82,32.24,17.82,10.12]","[15.41, 22.77, 29.92, 38.67, 42.17, 36.58, 44.96, 42.82, 32.24, 17.82, 10.12, 22.1]",5,Python,2025-08-14,"agentic-ai,embodied-ai,reinforcement-learning,rl-infra,rlinf,vla-rl",keep,Model Infra,Post-Train · RL & environments,"当前图中的 Post-Train · Reinforcement learning 代表项目；仓库仍在维护，4,289 stars，最近可用 OpenRank 17.52。",框架功能重叠明显，按通用性和持续协作控制在四个项目。,ok,,
975070903,llm-d/llm-d,Achieve state of the art inference performance with modern accelerators on Kubernetes,4649,216,790,223,Apache-2.0,false,2026-09-24,19.67,14.3,"[6.28,20.85,32.68,32.16,25.91,35.42,32.05,24.14,30.84,30.63,26.51,19.67]","[20.85, 32.68, 32.16, 25.91, 35.42, 32.05, 24.14, 30.84, 30.63, 26.51, 19.67, 14.3]",9,Shell,2025-04-29,"ai,cncf,distributed-inference,gpu,inference,intelligent-routing,kubernetes,llm,model-server",keep,Model Infra,Serving · Deploy,"当前图中的 Serving · Deploy 代表项目；仓库仍在维护，3,897 stars，最近可用 OpenRank 30.05。",该区只保留部署入口和 serving control plane，不与底层 inference engine 重复。,ok,,
1128123594,volcengine/OpenViking,"Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.",38613,281,3012,669,AGPL-3.0,false,2026-09-24,177.61,187.93,"[null,null,null,null,null,3.73,40.9,112.46,135.01,142.45,161.17,177.61]","[null, null, null, null, 3.73, 40.9, 112.46, 135.01, 142.45, 161.17, 177.61, 187.93]",163,Python,2026-01-05,"agent-memory,agent-plugins,agentic-rag,context-database,dsh-plugin,self-evolving",add,Agent Infra,"Memory, knowledge & context",把 memory、RAG 和 skills 收敛为 context database，补上当前图缺少的 agent-native context 数据层。,主仓库使用 AGPL-3.0；图例或许可证分析中需要明确标注。,ok,rising,2026-04 至 2026-07 OpenRank 135.01 → 142.45 → 161.17 → 177.61，连续上升。
974445592,microsoft/agent-framework,"A framework for building, orchestrating and deploying AI agents and multi-agent workflows with support for Python and .NET.",13776,267,2370,646,MIT,false,2026-09-24,27.23,28.47,"[4.65,10.74,90.35,113.71,111.77,101.41,91.06,78.76,66.29,54.08,44.9,27.23]","[10.74, 90.35, 113.71, 111.77, 101.41, 91.06, 78.76, 66.29, 54.08, 44.9, 27.23, 28.47]",20,Python,2025-04-28,"agent-framework,agentic-ai,agents,ai,dotnet,multi-agent,orchestration,python,sdk,workflows",add,Agent Infra,Code-first frameworks,统一 Python 与 .NET 的 agent 和 multi-agent 开发路径，补足企业开发栈与跨语言框架代表。,与 AutoGen、Semantic Kernel 有谱系关系；主图只保留当前统一框架，不并列堆放三个 Microsoft 项目。,ok,,
976095156,JetBrains/koog,"Koog is a JVM (Java and Kotlin) framework for building predictable, fault-tolerant and enterprise-ready AI agents across all platforms – from backend services to Android and iOS, JVM, and even in-browser environments. Koog is based on our AI products expertise and provides proven solutions for complex LLM and AI problems",4592,124,474,197,Apache-2.0,false,2026-09-24,,20.33,"[57.84,61.91,71.93,66.79,60.13,60.89,57.07,59.94,63.3,62.56,48.84,null]","[61.91, 71.93, 66.79, 60.13, 60.89, 57.07, 59.94, 63.3, 62.56, 48.84, null, 20.33]",1,Kotlin,2025-05-01,"agentframework,agentic-ai,agents,ai,ai-agents-framework,aiagentframework,android-ai,anthropic,genai,generative-ai,java,jvm,kotlin,ktor,llm,mcp,multi-agent-systems,ollama,openai,spring",add,Agent Infra,Code-first frameworks,为 Kotlin/JVM、Android 和 iOS 提供 agent framework，修正当前框架区过度集中在 Python、TypeScript 的问题。,社区规模仍小于 LangChain、Pydantic AI；入图理由是语言生态代表性，不是热度排名。,ok,,
925270205,trycua/cua,"Scale computer-use 2.0 with open-source drivers, cross-OS fleets, and benchmarks for training, evaluation, and data generation.",26223,110,1821,1057,MIT,false,2026-09-24,7.79,5.96,"[12.35,11.47,13.39,10.68,12.28,15.56,16.28,13.6,14.08,11.52,11.85,7.79]","[11.47, 13.39, 10.68, 12.28, 15.56, 16.28, 13.6, 14.08, 11.52, 11.85, 7.79, 5.96]",8,HTML,2025-01-31,"agent,ai-agent,apple,computer-use,computer-use-agent,containerization,cua,desktop-automation,hacktoberfest,lume,macos,manus,operator,swift,virtualization,virtualization-framework,windows,windows-sandbox",add,Agent Infra,"Tools, web & computer use",把 computer use 扩展到跨 OS driver、fleet、benchmark 和数据生成，结构价值高于再增加一个 browser agent。,最近可用 OpenRank 绝对值不高；保留依据主要是类别变化和当前关注度。,ok,,
950891299,agentgateway/agentgateway,Next Generation Agentic Proxy for AI Agents and MCP servers,5027,258,879,289,Apache-2.0,false,2026-09-24,17.51,14.89,"[16.6,19.73,17.27,20.0,19.49,18.09,22.94,27.51,27.15,24.54,22.96,17.51]","[19.73, 17.27, 20.0, 19.49, 18.09, 22.94, 27.51, 27.15, 24.54, 22.96, 17.51, 14.89]",8,Rust,2025-03-18,"agents,ai,ai-gateway,api-gateway,gateway-api,kubernetes,mcp,mcp-gateway,reverse-proxy,rust,service-mesh",add,Agent Infra,Protocols & interoperability,Agentic proxy 覆盖 MCP 与 agent 流量策略，归入 agent 协议与互操作层。,与 LiteLLM 的 model API gateway 定位不同，图中应直接标成 agentic proxy。,ok,,
979392131,ag-ui-protocol/ag-ui,AG-UI: the Agent-User Interaction Protocol. Bring Agents into Frontend Applications.,16024,149,1450,400,MIT,false,2026-09-24,5.38,3.47,"[22.5,22.14,25.54,28.04,26.73,23.46,27.79,20.29,18.85,13.41,11.03,5.38]","[22.14, 25.54, 28.04, 26.73, 23.46, 27.79, 20.29, 18.85, 13.41, 11.03, 5.38, 3.47]",11,TypeScript,2025-05-07,"ag-ui-protocol,agent-frontend,agent-ui,agentic-workflow,ai-agents",add,Agent Infra,Protocols & interoperability,补上 agent 到前端应用之间的事件与交互协议，当前协议区不再只围绕 MCP 和 A2A。,AG-UI 更偏交互事件和传输层；不要与 A2UI 合并成同一个概念。,ok,,
1063641074,a2ui-project/a2ui,,16503,93,1309,387,Apache-2.0,false,2026-09-24,33.04,28.24,"[null,0.16,8.76,11.12,46.81,48.43,40.48,49.15,51.59,54.66,49.66,33.04]","[0.16, 8.76, 11.12, 46.81, 48.43, 40.48, 49.15, 51.59, 54.66, 49.66, 33.04, 28.24]",26,TypeScript,2025-09-24,,add,Agent Infra,Protocols & interoperability,代表声明式 agent-generated UI，与 MCP、A2A、AG-UI 共同构成更完整的 agent interface 协议面。,官方仍标注 early-stage public preview，主图应加 preview 标识。,ok,,
1036441713,kubernetes-sigs/agent-sandbox,"agent-sandbox enables easy management of isolated, stateful, singleton workloads, ideal for use cases like AI agent runtimes  and reinforcement learning (RL).",4016,150,527,205,Apache-2.0,false,2026-09-24,11.35,6.5,"[1.31,11.37,15.81,20.36,22.21,20.29,21.95,22.86,32.85,22.26,18.96,11.35]","[11.37, 15.81, 20.36, 22.21, 20.29, 21.95, 22.86, 32.85, 22.26, 18.96, 11.35, 6.5]",14,Go,2025-08-12,"agent,ai,kubernetes,rl,sandbox",add,Agent Infra,Development sandboxes,Kubernetes SIG Apps 承载的声明式 sandbox orchestration，补上稳定身份、持久存储和 warm pool 这一层。,它是 sandbox orchestrator，不是底层隔离 runtime；与 OpenSandbox、Coder、Daytona 的角色需要区分。,ok,,
1054512829,vllm-project/vllm-omni,A framework for efficient model inference with omni-modality models,7051,448,1785,2107,Apache-2.0,false,2026-09-24,54.29,50.76,"[null,0.38,null,0.89,57.0,82.93,110.17,121.05,125.34,115.23,90.15,54.29]","[0.38, null, 0.89, 57.0, 82.93, 110.17, 121.05, 125.34, 115.23, 90.15, 54.29, 50.76]",30,Python,2025-09-11,"audio-generation,diffusion,image-generation,inference,model-serving,multimodal,pytorch,transformer,video-generation,world-model",add,Model Infra,Serving · Inference,把 vLLM serving 扩展到文本、图像、音频、视频和 action output，直接对应多模态 agent workload。,它与 vLLM 主项目同属一个生态；版面上应做子项目关系提示，避免看成两个无关引擎。,ok,,
807305060,LMCache/LMCache,LMCache: Supercharge Your LLM with the Fastest KV Cache Layer,11901,328,1951,755,Apache-2.0,false,2026-09-24,17.97,9.35,"[74.2,71.26,68.63,64.3,66.26,62.12,54.58,40.44,44.09,40.51,34.45,17.97]","[71.26, 68.63, 64.3, 66.26, 62.12, 54.58, 40.44, 44.09, 40.51, 34.45, 17.97, 9.35]",14,Python,2024-05-28,"amd,cuda,fast,inference,kv-cache,llm,pytorch,rocm,speed,vllm",add,Model Infra,Serving · Inference,KV cache 已成为长上下文和多轮 agent workload 的独立复用层；LMCache 的社区信号和采用度足以进入主图。,与 Mooncake 的位置高度重叠，本版采用替换而不是并列增加。,ok,,
924058625,vllm-project/vllm-ascend,Community maintained hardware plugin for vLLM on Huawei Ascend,2889,446,2365,3620,Apache-2.0,false,2026-09-24,48.15,44.74,"[139.78,150.66,133.79,157.1,186.32,182.38,143.35,144.3,127.4,96.91,73.88,48.15]","[150.66, 133.79, 157.1, 186.32, 182.38, 143.35, 144.3, 127.4, 96.91, 73.88, 48.15, 44.74]",60,Python,2025-01-29,"ascend,inference,llm,llm-serving,llmops,mlops,model-serving,transformer,vllm",omit,,,它是 vLLM 的 Ascend hardware plugin，适合做 vLLM 的硬件生态注释，不作为独立大 logo。,如果后续单独制作国产硬件适配图，再提升为主项目。,ok,,
250510075,huggingface/trl,Train transformer language models with reinforcement learning.,19379,465,3018,286,Apache-2.0,false,2026-09-24,22.75,19.59,"[88.54,80.62,74.66,68.59,62.79,53.86,54.41,40.69,40.63,34.95,27.31,22.75]","[80.62, 74.66, 68.59, 62.79, 53.86, 54.41, 40.69, 40.63, 34.95, 27.31, 22.75, 19.59]",14,Python,2020-03-27,,add,Model Infra,Post-Train · RL & environments,SFT、DPO、GRPO 等 post-training 的通用基础库，在 Hugging Face 生态中具有长期代表性。,项目不是 2026 新晋热点；这是旧图漏项修正，并替换社区信号较弱的 OpenRLHF。,ok,,
1166129534,NVIDIA/OpenShell,"OpenShell is the safe, private runtime for autonomous AI agents.",8777,122,1276,410,Apache-2.0,false,2026-09-24,30.27,27.36,"[null,null,null,null,null,null,0.26,14.12,22.15,33.98,32.81,30.27]","[null, null, null, null, null, 0.26, 14.12, 22.15, 33.98, 32.81, 30.27, 27.36]",17,Rust,2026-02-24,,omit,,,Sandbox 区加入 agent-sandbox 后已经有四个互补代表，OpenShell 与 NVIDIA agent runtime 产品栈重叠。,继续观察其外部贡献和独立社区结构。,ok,,
679343504,topoteretes/cognee,Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory across sessions with a self-hosted knowledge graph engine.,30959,282,3095,447,Apache-2.0,false,2026-09-24,19.93,15.03,"[30.35,34.89,34.63,35.59,32.1,27.23,31.4,27.86,27.21,19.8,22.28,19.93]","[34.89, 34.63, 35.59, 32.1, 27.23, 31.4, 27.86, 27.21, 19.8, 22.28, 19.93, 15.03]",22,Python,2023-08-16,"agent-memory,agent-skills,ai,ai-agents,ai-memory,cognitive-architecture,cognitive-memory,context-engineering,contributions-welcome,good-first-issue,good-first-pr,graph-database,graph-rag,help-wanted,knowledge,knowledge-graph,memory-management,open-source,vector-database",add,Agent Infra,"Memory, knowledge & context",长期记忆与 knowledge graph 结合，补足当前 memory 区对图结构和持久知识表达的覆盖。,与 RAGFlow、Mem0 有部分功能重叠；保留重点是 graph-based memory，不再额外增加同类项目。,ok,,
812044182,googleapis/mcp-toolbox,MCP Toolbox for Databases is an open source MCP server for databases.,16488,162,1732,327,Apache-2.0,false,2026-09-24,12.36,13.84,"[54.59,54.07,43.13,43.72,42.64,33.71,37.88,40.03,34.36,30.87,23.1,12.36]","[54.07, 43.13, 43.72, 42.64, 33.71, 37.88, 40.03, 34.36, 30.87, 23.1, 12.36, 13.84]",8,Go,2024-06-07,"agent,agents,ai,bigquery,clickhouse,cockroachdb,database,elasticsearch,firestore,genai,llm,mcp,mongodb,mysql,oracle,postgresql,redis,server,spanner,tidb",omit,,,数据库 MCP server 是重要集成案例，但不足以代表通用协议或 runtime。,适合作为 MCP 应用示例，不占协议主图 logo。,ok,,
1007406267,coze-dev/coze-loop,"Next-generation AI Agent Optimization Platform: Cozeloop addresses challenges in AI agent development by providing full-lifecycle management capabilities from development, debugging, and evaluation to monitoring.  ",5744,45,797,87,Apache-2.0,false,2026-09-24,30.54,28.95,"[25.61,30.45,30.98,33.33,34.36,32.87,34.03,31.85,33.97,35.61,31.47,30.54]","[30.45, 30.98, 33.33, 34.36, 32.87, 34.03, 31.85, 33.97, 35.61, 31.47, 30.54, 28.95]",16,Go,2025-06-24,"agent,agent-evaluation,agent-observability,agentops,ai,coze,eino,evaluation,langchain,llm-observability,llmops,monitoring,observability,open-source,openai,playground,prompt-management",omit,,,Observability 区已有 Opik、Langfuse、Phoenix 和 Promptfoo，当前无需增加第五个相似平台。,若外部采用和独立贡献者继续上升，可进入替换候选。,ok,,
1004147641,microsoft/agent-lightning,The absolute trainer to light up AI agents.,18464,44,1634,160,MIT,false,2026-09-17,,2.45,"[10.35,9.99,15.41,19.07,24.52,21.59,14.87,5.31,1.58,0.95,0.71,null]","[9.99, 15.41, 19.07, 24.52, 21.59, 14.87, 5.31, 1.58, 0.95, 0.71, null, 2.45]",1,Python,2025-06-18,"agent,agentic-ai,llm,mlops,reinforcement-learning",omit,,,面向 agent training 的方向值得跟踪，但最近可用 OpenRank 仅 0.62，暂不足以进入 Post-Train 主图。,等持续协作信号稳定后再复核。,ok,,
790916393,NVIDIA/Model-Optimizer,"A unified library of SOTA model optimization techniques like quantization, distillation, pruning, neural architecture search, speculative decoding, etc. It compresses deep learning models for downstream deployment frameworks like TensorRT-LLM, TensorRT, vLLM, etc. to optimize inference speed.",3936,90,624,417,Apache-2.0,false,2026-09-24,36.92,36.21,"[11.46,29.36,38.47,44.84,47.68,53.42,55.52,57.77,56.85,52.51,49.88,36.92]","[29.36, 38.47, 44.84, 47.68, 53.42, 55.52, 57.77, 56.85, 52.51, 49.88, 36.92, 36.21]",22,Python,2024-04-23,,add,Model Infra,Pre-Train · Compiler & accelerator,统一量化、蒸馏、剪枝和投机解码等部署优化能力，最近可用 OpenRank 56.14。,厂商主导且与 TensorRT-LLM 有交叉；主图只表达模型优化层，不扩展 NVIDIA 同栈项目数量。,ok,,
949546994,NVIDIA-NeMo/RL,Scalable toolkit for efficient model reinforcement,2031,163,571,1134,Apache-2.0,false,2026-09-24,35.74,34.1,"[61.83,64.77,65.51,48.69,47.48,46.68,38.04,42.5,42.08,50.07,45.3,35.74]","[64.77, 65.51, 48.69, 47.48, 46.68, 38.04, 42.5, 42.08, 50.07, 45.3, 35.74, 34.1]",26,Python,2025-03-16,,add,Model Infra,Post-Train · RL & environments,OpenRank 35.74、7 月参与者 26，补足成熟的分布式 post-training 工程实现。,NVIDIA 在 Model Infra 已有多个项目；这里表达独立的 post-training 角色，不外推为采用结论。,ok,,
1157171418,jundot/omlx,LLM inference server with continuous batching & SSD caching for Apple Silicon — managed from the macOS menu bar,22179,270,1922,1428,Apache-2.0,false,2026-09-24,12.13,12.29,"[null,null,null,null,null,null,5.15,19.25,35.66,35.83,26.17,12.13]","[null, null, null, null, null, 5.15, 19.25, 35.66, 35.83, 26.17, 12.13, 12.29]",12,Python,2026-02-13,"apple-silicon,inference-server,llm,macos,mlx,openai-api",omit,,,Apple Silicon inference 很有关注度，但单平台项目暂不挤占已经扩容的 Serving · Inference 区。,观察其持续贡献和非 macOS 扩展情况。,ok,,
947349598,stacklok/toolhive,ToolHive is an enterprise-grade platform for running and managing Model Context Protocol (MCP) servers.,2205,131,305,372,Apache-2.0,false,2026-09-23,16.71,14.53,"[39.55,37.58,32.56,33.14,32.49,30.9,31.52,33.65,32.84,31.23,16.23,16.71]","[37.58, 32.56, 33.14, 32.49, 30.9, 31.52, 33.65, 32.84, 31.23, 16.23, 16.71, 14.53]",11,Go,2025-03-12,"ai,ai-security,aicodeassistant,golang,kubernetes,mcp,mcp-security,mcp-servers,mcp-tools,model-context-protocol,security",add,Agent Infra,Protocols & interoperability,提供 MCP server runtime、registry、policy、隔离和 Kubernetes 运维，补足协议落地后的控制面。,当前社区规模小于头部 gateway 项目；入图依据是独立的运行与治理角色。,ok,,
1043426082,OpenHands/software-agent-sdk,"A clean, modular SDK for building AI agents with OpenHands V1.",1165,125,558,549,MIT,false,2026-09-24,36.33,26.16,"[4.03,32.13,46.32,52.89,46.9,46.54,49.73,50.37,49.92,54.75,68.82,36.33]","[32.13, 46.32, 52.89, 46.9, 46.54, 49.73, 50.37, 49.92, 54.75, 68.82, 36.33, 26.16]",10,Python,2025-08-23,"agent,sdk",omit,,,它是 OpenHands 的拆分 SDK；主图已经保留 OpenHands，不重复增加同一项目谱系 logo。,在 OpenHands 项目演化说明中标注即可。,ok,,
1085876386,microsoft/fara,Fara1.5 – A family of frontier computer use agent models,6192,7,603,42,MIT,false,2026-09-22,,0.86,"[null,null,null,5.19,6.06,2.9,2.18,null,3.3,null,null,null]","[null, null, 5.19, 6.06, 2.9, 2.18, null, 3.3, null, null, null, 0.86]",0,Python,2025-10-29,"agent,browser-use,computer-use,computer-use-agent,cua",omit,,,Fara 属于 computer-use model 候选，Large Models 需要按模型发布、开放权重和评测单独筛选。,不能直接沿用 GitHub repo 的 OpenRank 规则决定入图。,ok,,
946829547,NVIDIA/Isaac-GR00T,NVIDIA Isaac GR00T N1.7 -  A Foundation Model for Generalist Robots.,8121,36,1469,337,Apache-2.0,false,2026-08-20,,1.05,"[27.9,23.75,16.71,12.53,10.68,8.97,8.51,4.73,4.67,1.67,null,null]","[23.75, 16.71, 12.53, 10.68, 8.97, 8.51, 4.73, 4.67, 1.67, null, null, 1.05]",0,Python,2025-03-11,,omit,,,Isaac GR00T 属于 robotics foundation model，Large Models 使用独立发布与开放度口径。,本 CSV 保留项目数据，但不把它直接塞入 Agent Infra 或 Model Infra。,ok,,
1166102148,DeusData/codebase-memory-mcp,"High-performance code intelligence MCP server. Indexes codebases into a persistent knowledge graph — average repo in milliseconds. 158 languages, sub-ms queries, 99% fewer tokens. Single static binary, zero dependencies.",44806,159,3656,606,MIT,false,2026-09-24,8.74,7.48,"[null,null,null,null,null,null,1.18,9.33,7.58,7.39,8.67,8.74]","[null, null, null, null, null, 1.18, 9.33, 7.58, 7.39, 8.67, 8.74, 7.48]",11,C,2026-02-24,"aider,ast,claude-code,code-analysis,code-intelligence,codex,cursor,cypher,developer-tools,gemini-cli,graph-visualization,kilocode,knowledge-graph,mcp,mcp-server,model-context-protocol,opencode,sqlite,tree-sitter,windsurf",remove,Agent Infra,"Memory, knowledge & context",同类代码知识图谱项目集中涌现，Memory/context 分区需要控制重复。,保留在观察池，与 Graphify、CodeGraph 等同类持续比较。,ok,,
1152451571,withastro/flue,The sandbox agent framework.,8364,22,500,39,Apache-2.0,false,2026-09-24,8.83,2.57,"[null,null,null,null,null,null,0.61,0.85,2.65,12.5,9.91,8.83]","[null, null, null, null, null, 0.61, 0.85, 2.65, 12.5, 9.91, 8.83, 2.57]",3,TypeScript,2026-02-07,,remove,Agent Infra,Code-first frameworks,当前 OpenRank 8.83、7 月参与者 3；Code-first frameworks 已有更活跃的代表项目。,保留在完整项目池中观察；本次退图是密度控制，不等于否定项目价值。,ok,,
1183888342,stablyai/orca,"Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.",77253,385,5064,6734,MIT,false,2026-09-24,29.1,33.25,"[null,null,null,null,null,null,null,4.24,13.86,21.31,26.42,29.1]","[null, null, null, null, null, null, 4.24, 13.86, 21.31, 26.42, 29.1, 33.25]",49,TypeScript,2026-03-17,"ade,agent-ide,ai-agents,claude-code,cli,codex,cursor-agent,devtools,ghostty,ide,mobile-app,opencode,orchestration,parallel-agents,pi,terminal,worktrees,yc-backed",add,Agent Infra,Multi-agent orchestration,为并行 coding agents 提供桌面、移动端和远程控制面；连续 5 周上榜，且 2026-07 有 49 位 issue/PR 参与者，形成了现有编排层之外的新产品形态。,项目创建于 2026 年，仍需观察长期留存和团队采用。,ok,rising,2026-04 至 2026-07 OpenRank 13.86 → 21.31 → 26.42 → 29.10，连续上升。
1129940957,headroomlabs-ai/headroom,"Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.",73697,269,5688,669,Apache-2.0,false,2026-09-24,13.92,9.96,"[null,null,null,null,null,2.1,2.43,3.05,9.58,9.48,15.91,13.92]","[null, null, null, null, 2.1, 2.43, 3.05, 9.58, 9.48, 15.91, 13.92, 9.96]",18,Python,2026-01-07,"agent,ai,anthropic,claude-code,compression,context-engineering,context-window,cursor,fastapi,langchain,llm,mcp,openai,prompt-engineering,proxy,python,rag,token-optimization,tokens,typescript",add,Agent Infra,"Memory, knowledge & context",压缩工具输出、日志和 RAG 内容以降低 agent 上下文成本，兼具 library、proxy 和 MCP server 形态；连续 3 周上榜。,README 中的压缩比例是项目方测试结果，需要独立 benchmark 复核。,ok,,
1241960226,alibaba/open-code-review,"Secure, fast, efficient, battle-tested at Alibaba's scale. Hybrid architecture code review tool: deterministic pipelines + LLM Agent, precise line-level comments, built-in multi-language ruleset (NPE, thread-safety, XSS, SQL injection), OpenAI & Anthropic compatible.",40577,192,2921,216,Apache-2.0,false,2026-09-24,2.96,8.5,"[null,null,null,null,null,null,null,null,null,0.34,3.25,2.96]","[null, null, null, null, null, null, null, null, 0.34, 3.25, 2.96, 8.5]",9,Go,2026-05-18,"agent,agent-skills,code-review,code-review-assistant,harness,repository-level-context",remove,Agent Infra,Agentic coding,定位清晰，但当前 OpenRank 2.96，外部采用和 benchmark 仍待验证。,项目仍保留在完整项目池，后续按持续协作复核。,ok,new,NEW：created_at 2026-05-18；按 2026-08 月度快照口径，创建日期在 2026-05-01 及以后。
1216785679,esengine/DeepSeek-Reasonix,DeepSeek-native AI coding agent for your terminal. Engineered around prefix-cache stability — leave it running.,35704,153,2421,1915,MIT,false,2026-09-24,26.06,22.79,"[null,null,null,null,null,null,null,null,1.6,30.86,34.01,26.06]","[null, null, null, null, null, null, null, 1.6, 30.86, 34.01, 26.06, 22.79]",28,Go,2026-04-21,"agent,agent-framework,ai-agent,ai-coding,cli,coding-agent,deepseek,developer-tools,dsh,dsh-plugin,ink,llm,prompt-caching,r1,terminal,tool-use,tui,typescript",add,Agent Infra,Agentic coding,"围绕 DeepSeek 和 prefix-cache 稳定性设计的终端 coding agent；34,916 stars、OpenRank 34.01、7 月参与者 28，体现了独立技术路线。",模型绑定较强，需要观察能否形成多模型或更通用的使用面。,ok,,
1157102282,diegosouzapw/OmniRoute,"Never stop coding. Free MIT AI gateway: one endpoint, 359 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by hundreds of contributors",69814,431,9906,545,MIT,false,2026-09-24,31.92,21.82,"[null,null,null,null,null,null,4.48,23.7,32.18,41.54,39.17,31.92]","[null, null, null, null, null, 4.48, 23.7, 32.18, 41.54, 39.17, 31.92, 21.82]",30,TypeScript,2026-02-13,"a2a,ai-agents,ai-gateway,anthropic,claude,claude-code,cline,codex,copilot,cursor,deepseek,free-ai,gemini,kimi,llm-gateway,mcp,openai,openai-proxy,qwen,token-saver",add,Model Infra,Model API gateways,提供多 provider、多模型的统一网关、配额感知 fallback 和协议适配；连续 4 周上榜，OpenRank 39.17、7 月参与者 30。,增长很快且功能声明较多，需要核验 provider 覆盖、许可证依赖和生产稳定性。,ok,,
1246459259,MoonshotAI/kimi-code,Kimi Code CLI  —  The Starting Point for Next-Gen Agents,7659,56,1245,1464,MIT,false,2026-09-24,19.27,27.34,"[null,null,null,null,null,null,null,null,null,12.26,27.5,19.27]","[null, null, null, null, null, null, null, null, 12.26, 27.5, 19.27, 27.34]",13,TypeScript,2026-05-22,,add,Agent Infra,Agentic coding,MoonshotAI 官方新一代 Kimi Code CLI，OpenRank 27.5、7 月参与者 13，具备明确的 coding-agent 产品形态。,与数据集中已有但暂未入图的 MoonshotAI/kimi-cli 存在重叠，需确认长期仓库关系后再决定主图替换。,ok,new,NEW：created_at 2026-05-22；按 2026-08 月度快照口径，创建日期在 2026-05-01 及以后。 5—7 月 OpenRank 12.26 / 27.50 / 19.27。
1193909050,herdrdev/herdr,the runtime your coding agents live on,40553,87,3099,359,Apache-2.0,false,2026-09-24,11.23,13.14,"[null,null,null,null,null,null,null,1.16,4.45,13.15,11.89,11.23]","[null, null, null, null, null, null, 1.16, 4.45, 13.15, 11.89, 11.23, 13.14]",9,Rust,2026-03-27,"agent,agent-orchestration,ai,ai-agents,claude-code,cli,codex,coding-agents,developer-tools,devtools,multiplexer,rust,terminal,terminal-multiplexer,terminal-ui,tmux,tui,workspace-manager",add,Agent Infra,Coding workflows & harnesses,面向多种 coding agents 的运行时和终端 multiplexer，连续 3 周上榜，补足 agent workspace/runtime 这一层。,与 Orca、T3 Code 的控制面有交叉，后续应按运行时能力而非 UI 热度比较。,ok,,
1187630119,NVIDIA/SkillSpector,"Security scanner for AI agent skills. Detect vulnerabilities, malicious patterns, security risks, prompt injection, data exfiltration, and supply-chain risks in Claude Code, Codex, and MCP skills before you install them.",18205,79,1573,123,Apache-2.0,false,2026-09-24,1.93,9.22,"[null,null,null,null,null,null,null,null,null,null,3.44,1.93]","[null, null, null, null, null, null, null, null, null, 3.44, 1.93, 9.22]",4,Python,2026-03-21,"agent-security,agent-skills,agentic-ai,ai-security,claude-code,mcp,prompt-injection,security-scanner,security-tools,security-workflow,supply-chain-security",remove,Agent Infra,Observability & evaluation,Agent security 是值得跟踪的方向，但当前没有合适分类，OpenRank 1.93、参与者 4。,等至少三个持续活跃的安全项目形成稳定分区后再拆类。,ok,,
1153130349,pingdotgg/t3code,,23466,370,6055,2233,MIT,false,2026-09-24,44.99,49.56,"[null,null,null,null,null,null,null,90.0,86.42,67.84,47.98,44.99]","[null, null, null, null, null, null, 90.0, 86.42, 67.84, 47.98, 44.99, 49.56]",28,TypeScript,2026-02-08,,add,Agent Infra,Coding workflows & harnesses,为 Codex、Claude Code、Cursor、OpenCode 等本地 agent 提供桌面、Web 和移动控制面；OpenRank 47.98、7 月参与者 28。,README 明确说明项目仍很早期，当前社区贡献政策也较收敛。,ok,,
1133911335,different-ai/openwork,The open-source alternative to Claude Cowork (powered by opencode),23717,69,2385,543,NOASSERTION,false,2026-09-24,11.13,10.06,"[null,null,null,null,null,12.59,15.24,21.64,20.56,19.03,13.91,11.13]","[null, null, null, null, 12.59, 15.24, 21.64, 20.56, 19.03, 13.91, 11.13, 10.06]",6,TypeScript,2026-01-14,,remove,Agent Infra,Workflow & agent builders,当前 OpenRank 11.13、参与者 6，且许可证字段为 NOASSERTION；builder 分区已经较拥挤。,完成许可边界和持续性复核后可重新进入。,ok,,
764287615,supermemoryai/supermemory,"Memory and context engine + app that is extremely fast, scalable, and can be run fully locally. The Memory API for the AI era.",30873,106,2704,120,MIT,false,2026-09-24,6.7,6.06,"[5.4,6.37,13.6,11.92,9.69,12.12,10.12,9.02,7.27,8.61,11.26,6.7]","[6.37, 13.6, 11.92, 9.69, 12.12, 10.12, 9.02, 7.27, 8.61, 11.26, 6.7, 6.06]",9,TypeScript,2024-02-27,"agent-memory,ai-memory,cloudflare-kv,cloudflare-pages,cloudflare-workers,drizzle-orm,memory,postgres,remix,tailwindcss,typescript,vite",add,Agent Infra,"Memory, knowledge & context","同时提供本地 memory/context engine、API 和 MCP 接入，29,000 左右 stars 且 7 月仍有持续协作。",产品与开源引擎边界较宽，应避免把项目方 benchmark 直接写成中立结论。,ok,,
1067949357,huggingface/OpenEnv,An interface library for RL post training with environments. ,2606,92,460,106,BSD-3-Clause,false,2026-09-24,,6.65,"[null,null,10.91,22.0,18.86,18.32,15.03,11.32,9.36,8.51,7.55,null]","[null, 10.91, 22.0, 18.86, 18.32, 15.03, 11.32, 9.36, 8.51, 7.55, null, 6.65]",6,Python,2025-10-01,,add,Model Infra,Post-Train · RL & environments,Hugging Face 提供的 RL post-training environment 接口，直接连接 agentic execution environments 与训练流程。,当前规模较小，仅上榜 1 周；需要观察环境生态和训练框架接入。,ok,,
888092115,microsoft/markitdown,Python tool for converting files and office documents to Markdown.,186768,128,13777,695,MIT,false,2026-09-21,2.08,5.25,"[13.93,10.36,7.05,5.91,5.98,5.34,6.79,6.01,10.96,7.8,5.08,2.08]","[10.36, 7.05, 5.91, 5.98, 5.34, 6.79, 6.01, 10.96, 7.8, 5.08, 2.08, 5.25]",11,Python,2024-11-13,"autogen,autogen-extension,langchain,markdown,microsoft-office,openai,pdf",add,Model Infra,Data · Integration,"将 Office、PDF 等文件统一转换为 Markdown，连续 3 周上榜且达到 174,843 stars，可作为 agent/RAG 数据入口基础工具。",它是通用文档转换器，Agentic AI 价值来自上游数据准备，不应描述成 agent framework。,ok,,
858798469,allenai/olmocr,Toolkit for linearizing PDFs for LLM datasets/training,19656,15,1641,90,Apache-2.0,false,2026-03-25,,,"[15.38,12.0,10.27,10.84,7.59,4.35,4.33,2.82,1.36,1.78,null,null]","[12.0, 10.27, 10.84, 7.59, 4.35, 4.33, 2.82, 1.36, 1.78, null, null, null]",0,Python,2024-09-17,,remove,Model Infra,Data · Integration,2026-07 OpenRank 缺失、参与者为 0；Data · Integration 已有 Docling 和 MarkItDown。,继续作为模型数据准备候选观察。,ok,,
1200597263,Graphify-Labs/graphify,"Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.",121123,260,11678,1459,Apache-2.0,false,2026-09-23,13.27,7.14,"[null,null,null,null,null,null,null,null,22.48,21.04,17.65,13.27]","[null, null, null, null, null, null, null, 22.48, 21.04, 17.65, 13.27, 7.14]",11,Python,2026-04-03,"ai-agents,antigravity,ast,claude-code,code-analysis,code-search,codex,cursor,developer-tools,gemini,graphrag,knowledge-graph,leiden,llm,mcp,openclaw,rag,skills,tree-sitter",remove,Agent Infra,"Memory, knowledge & context",与 codebase-memory-mcp 等代码知识图谱项目高度重叠，项目历史也较短。,保留在观察池，等待独立采用和持续协作证据。,ok,,
1333065091,deepseek-ai/deepseek-harness,DeepSeek Harness: Everything is a Plugin.,234965,41,28264,0,MIT,false,2026-09-24,,,"[null,null,null,null,null,null,null,null,null,null,null,null]","[null, null, null, null, null, null, null, null, null, null, null, null]",0,TypeScript,2026-08-13,"ai-agents,cordis,dsh,dsh-plugin",add,Agent Infra,Agentic coding,DeepSeek AI 官方 agent harness，以插件架构提供可扩展的 coding-agent 运行与工具体系；发布期 Stars 增长显著。,项目仍处于 developer preview；RISING 只描述发布期关注增长，不表示社区成熟度或生产采用。,ok,new+rising,NEW：created_at 2026-08-13；按 2026-08 月度快照口径，创建日期在 2026-05-01 及以后。 RISING：发布期 Stars 快速累积；该标记不代表社区成熟度或生产采用。
1256070689,MiniMax-AI/minimax-code,"An open-source coding agent for your terminal, powered by MiniMax.",1818,12,209,122,MIT,false,2026-09-24,,,"[null,null,null,null,null,null,null,null,null,null,null,null]","[null, null, null, null, null, null, null, null, null, null, null, null]",0,TypeScript,2026-06-01,"coding-agents,harness,minimax",add,Agent Infra,Agentic coding,MiniMax 官方开源的终端 coding agent，直接面向代码生成与仓库内开发任务；补充 MiniMax 在 Agentic coding 区的开源项目代表。,项目创建时间较短，2026-07 OpenRank 尚无公开快照；NEW 只表示近期创建，不代表社区成熟度或生产采用。,ok,new,NEW：created_at 2026-06-01；按 2026-08 月度快照口径，创建日期在 2026-05-01 及以后。
1042367133,github/spec-kit,💫 Toolkit to help you get started with Spec-Driven Development,138717,295,12427,292,MIT,false,2026-09-24,15.58,10.61,"[null,96.31,69.26,39.04,26.2,21.62,16.09,17.89,22.53,20.54,19.84,15.58]","[96.31, 69.26, 39.04, 26.2, 21.62, 16.09, 17.89, 22.53, 20.54, 19.84, 15.58, 10.61]",13,Python,2025-08-21,"ai,copilot,development,engineering,prd,spec,spec-driven",add,Agent Infra,Coding workflows & harnesses,把 specification、plan 和 tasks 做成可由不同 coding agent 执行的流程，补上代码生成之前的公共工作层。,关注度很高，但 OpenRank 已从发布期峰值回落；入图表达 spec-driven workflow，不解释成近期协作增长。,ok,,
1167965563,openai/symphony,"Symphony turns project work into isolated, autonomous implementation runs, allowing teams to manage work instead of supervising coding agents.",27394,9,2837,8,Apache-2.0,false,2026-09-15,,,"[null,null,null,null,null,null,null,0.12,0.43,0.34,0.53,null]","[null, null, null, null, null, null, 0.12, 0.43, 0.34, 0.53, null, null]",1,Elixir,2026-02-26,,remove,Agent Infra,Coding workflows & harnesses,仍处于 engineering preview，2026-07 OpenRank 缺失、参与者为 1。,等公开协作和长期维护信号稳定后再评估。,ok,,
1191372811,larksuite/cli,"The official Lark/飞书 CLI tool, maintained by the larksuite team — built for humans and AI Agents. Covers core business domains including Messenger, Docs, Base, Sheets, Calendar, Mail, Tasks, Meetings, and more, with 200+ commands and 20+ AI Agent Skills.",17446,165,1395,697,MIT,false,2026-09-24,179.37,186.94,"[null,null,null,null,null,null,null,27.1,95.47,138.25,168.45,179.37]","[null, null, null, null, null, null, 27.1, 95.47, 138.25, 168.45, 179.37, 186.94]",190,Go,2026-03-25,,add,Agent Infra,"Tools, web & computer use",把文档、日历、邮件和消息等 SaaS 能力交付成结构化 CLI 与 Agent Skills，代表业务软件正在提供 agent-native 入口。,能力集中在 Lark/飞书生态；入图表达工具入口的变化，不把单一厂商覆盖面外推到整个 SaaS 市场。,ok,rising,2026-04 至 2026-07 OpenRank 95.47 → 138.25 → 168.45 → 179.37，连续上升。
1232669890,microsoft/SkillOpt,"SkillOpt is a text-space optimizer that trains reusable natural-language skills for frozen LLM agents through trajectory-driven edits, validation-gated updates, and deployable best_skill.md artifacts.",17432,59,1630,53,MIT,false,2026-09-05,1.56,4.6,"[null,null,null,null,null,null,null,null,null,1.74,2.22,1.56]","[null, null, null, null, null, null, null, null, 1.74, 2.22, 1.56, 4.6]",2,Python,2026-05-08,"agent-skills,self-evolving-agents",remove,Agent Infra,Observability & evaluation,当前 OpenRank 1.56、参与者 2，且 skill 优化不等同于 observability/evaluation。,保留为 agent improvement 候选；未来形成独立类别时再复核。,ok,new,NEW：created_at 2026-05-08；按 2026-08 月度快照口径，创建日期在 2026-05-01 及以后。
787076358,firecrawl/firecrawl,"The web data API to search, scrape, and interact at scale. 🔥",184144,166,9908,646,AGPL-3.0,false,2026-09-24,5.72,7.0,"[36.13,40.18,34.37,34.13,36.3,28.75,24.82,21.13,13.29,8.05,5.93,5.72]","[40.18, 34.37, 34.13, 36.3, 28.75, 24.82, 21.13, 13.29, 8.05, 5.93, 5.72, 7.0]",10,TypeScript,2024-04-15,"ai,ai-agents,ai-crawler,ai-scraping,ai-search,crawler,data-extraction,html-to-markdown,llm,markdown,scraper,scraping,web-crawler,web-data,web-data-extraction,web-scraper,web-scraping,web-search,webscraping",add,Agent Infra,"Tools, web & computer use",覆盖 search、scrape、crawl 与页面交互，把开放 Web 变成 agent 可调用的上下文接口；这是上一轮按近期增速筛选时漏掉的成熟基础设施。,Stars 很高但近期 OpenRank 下行；社区版采用 AGPL-3.0，托管服务和自部署边界需要分开理解。,ok,,
1266212515,omnigent-ai/omnigent,"Omnigent is an open-source AI agent framework and meta-harness: orchestrate Claude Code, Codex, Cursor, Pi, and custom agents — swap harnesses without rewriting, enforce policies and sandboxing, and collaborate in real time from any device.",10203,262,1620,1446,Apache-2.0,false,2026-09-24,32.65,31.14,"[null,null,null,null,null,null,null,null,null,null,29.98,32.65]","[null, null, null, null, null, null, null, null, null, 29.98, 32.65, 31.14]",24,Python,2026-06-11,"agent-framework,agent-governance,agent-orchestration,agents,ai,ai-agent,ai-agents,claude-code,codex,coding-agents,developer-tools,llm,ml,multi-agent,python,sandbox",omit,,,6 月创建后，7 月 OpenRank 升到 32.65；meta-harness 与 agent governance 值得继续跟踪。,Coding harnesses 与 framework 两区已经很满，先观察外部采用和持续协作，再决定替换谁。,ok,new,NEW：created_at 2026-06-11；按 2026-08 月度快照口径，创建日期在 2026-05-01 及以后。 6—7 月 OpenRank 29.98 → 32.65。
1300902838,xai-org/grok-build,"SpaceXAI's coding agent harness and TUI. Fullscreen, mouse interactive, extensible.",27070,1,5104,0,Apache-2.0,false,2026-09-23,,,"[null,null,null,null,null,null,null,null,null,null,null,null]","[null, null, null, null, null, null, null, null, null, null, null, null]",0,Rust,2026-07-14,,omit,,,7 月中旬发布后快速进入 GitHub Trending，是本轮必须进入观察池的 coding harness。,项目历史不足三周，7 月 OpenRank 尚无有效值；先不挤入已经拥挤的 Agentic coding 主图。,ok,new,NEW：created_at 2026-07-14；按 2026-08 月度快照口径，创建日期在 2026-05-01 及以后。 短期 attention 很强，持续协作尚待验证。
1271113810,vercel/eve,The Open Framework for Building Agents,5347,60,588,920,Apache-2.0,false,2026-09-24,12.88,21.78,"[null,null,null,null,null,null,null,null,null,null,5.68,12.88]","[null, null, null, null, null, null, null, null, null, 5.68, 12.88, 21.78]",16,TypeScript,2026-06-16,"agent,framework,harness,javascript,markdown,sandbox,typescript,vercel,workflows",omit,,,以 markdown、sandbox 和 workflow 组织 agent framework，7 月 OpenRank 升到 12.88。,与 Vercel AI SDK 及现有 code-first frameworks 有谱系重叠，先保留为替换候选。,ok,new,NEW：created_at 2026-06-16；按 2026-08 月度快照口径，创建日期在 2026-05-01 及以后。 6—7 月 OpenRank 5.68 → 12.88。
1378288598,zai-org/ZCode,"Z.ai's coding agent harness. Powerful, intelligent, extensible.",6822,2,2050,11,Apache-2.0,false,2026-09-24,,,"[null,null,null,null,null,null,null,null,null,null,null,null]","[null, null, null, null, null, null, null, null, null, null, null, null]",0,TypeScript,2026-09-20,,add,Agent Infra,Agentic coding,Z.ai 官方开源的 coding agent harness，覆盖桌面、浏览器和终端形态，补充 Agentic coding 区的新项目代表。,仓库于 2026-09-20 创建，8 月 OpenRank 尚无数据；NEW 只表示近期创建，不代表社区成熟度或生产采用。,ok,new,NEW：created_at 2026-09-20；创建时间晚于 2026-08 月度快照，OpenRank 留空。