[{"data":1,"prerenderedAt":1086},["ShallowReactive",2],{"doc-page:\u002Fdocs\u002Fai-image-gen":3},{"doc":4,"prev":1048,"next":1052,"resolvedType":8,"readingMinutes":498,"audience":1059,"checklist":1063,"related":1067},{"path":5,"title":6,"description":7,"docType":8,"resourceKind":9,"categoryId":10,"categoryLabel":11,"updatedAt":12,"publishedAt":12,"icon":13,"body":14},"\u002Fdocs\u002Fai-image-gen","AI 图像生成","Stable Diffusion、ComfyUI、Civitai 模型与本地部署指南","article",null,"ai-tools","AI 工具","2026-02-27","i-carbon-chat-bot",{"type":15,"value":16,"toc":1009},"minimark",[17,21,25,29,64,67,226,230,296,299,302,310,313,316,327,330,333,337,340,385,388,413,417,453,456,460,463,515,521,525,528,571,576,579,590,593,596,614,617,620,623,626,636,640,643,652,656,683,686,757,760,763,766,780,784,787,801,804,807,821,824,838,841,861,864,867,875,878,884,887,947,950,1005],[18,19,6],"h1",{"id":20},"ai-图像生成",[22,23,24],"p",{},"这页适合作为“AI 出图工具与本地工作流入口”。想把图像生成真正用顺，通常要同时考虑平台选择、模型类型、提示词结构、工作流工具和显存预算，而不是只盯着单个模型名。",[26,27,28],"h2",{"id":28},"先按目标选",[30,31,32,40,46,52,58],"ul",{},[33,34,35,39],"li",{},[36,37,38],"strong",{},"想要最快出效果","：优先在线平台",[33,41,42,45],{},[36,43,44],{},"想做长期迭代与批量工作流","：优先 ComfyUI",[33,47,48,51],{},[36,49,50],{},"想要传统一站式界面","：优先 Stable Diffusion WebUI",[33,53,54,57],{},[36,55,56],{},"想做文字排版、海报、Logo","：优先 Ideogram、Recraft 这类强文字与设计平台",[33,59,60,63],{},[36,61,62],{},"想要高度可控的本地生成","：优先本地模型 + 节点工作流",[26,65,66],{"id":66},"在线平台",[68,69,70,86],"table",{},[71,72,73],"thead",{},[74,75,76,80,83],"tr",{},[77,78,79],"th",{},"平台",[77,81,82],{},"说明",[77,84,85],{},"定价",[87,88,89,107,122,137,152,167,182,197,211],"tbody",{},[74,90,91,101,104],{},[92,93,94],"td",{},[95,96,100],"a",{"href":97,"rel":98},"https:\u002F\u002Fwww.midjourney.com\u002F",[99],"nofollow","Midjourney",[92,102,103],{},"艺术风格最强，社区活跃",[92,105,106],{},"$10\u002F月起",[74,108,109,116,119],{},[92,110,111],{},[95,112,115],{"href":113,"rel":114},"https:\u002F\u002Fopenai.com\u002Fdall-e-3",[99],"DALL-E 3",[92,117,118],{},"ChatGPT 内置，易用性最佳",[92,120,121],{},"ChatGPT Plus $20\u002F月",[74,123,124,131,134],{},[92,125,126],{},[95,127,130],{"href":128,"rel":129},"https:\u002F\u002Fideogram.ai\u002F",[99],"Ideogram",[92,132,133],{},"文字渲染能力最强",[92,135,136],{},"免费 \u002F $8\u002F月起",[74,138,139,146,149],{},[92,140,141],{},[95,142,145],{"href":143,"rel":144},"https:\u002F\u002Fwww.recraft.ai\u002F",[99],"Recraft",[92,147,148],{},"设计师向，Logo\u002F图标\u002FUI 素材",[92,150,151],{},"免费 \u002F $25\u002F月",[74,153,154,161,164],{},[92,155,156],{},[95,157,160],{"href":158,"rel":159},"https:\u002F\u002Fleonardo.ai\u002F",[99],"Leonardo AI",[92,162,163],{},"游戏角色和概念艺术",[92,165,166],{},"免费 \u002F $12\u002F月起",[74,168,169,176,179],{},[92,170,171],{},[95,172,175],{"href":173,"rel":174},"https:\u002F\u002Ffirefly.adobe.com\u002F",[99],"Adobe Firefly",[92,177,178],{},"Adobe 生态集成，商用安全",[92,180,181],{},"Creative Cloud 订阅",[74,183,184,191,194],{},[92,185,186],{},[95,187,190],{"href":188,"rel":189},"https:\u002F\u002Fdesigner.microsoft.com\u002F",[99],"Microsoft Designer",[92,192,193],{},"微软 AI 设计，免费",[92,195,196],{},"免费",[74,198,199,206,209],{},[92,200,201],{},[95,202,205],{"href":203,"rel":204},"https:\u002F\u002Fstitch.withgoogle.com\u002F",[99],"Stitch",[92,207,208],{},"Google AI 设计工具",[92,210,196],{},[74,212,213,220,223],{},[92,214,215],{},[95,216,219],{"href":217,"rel":218},"https:\u002F\u002Fcivitai.com\u002F",[99],"Civitai",[92,221,222],{},"AI 模型社区，可在线生成",[92,224,225],{},"免费 \u002F 按量",[227,228,229],"h3",{"id":229},"选择建议",[68,231,232,242],{},[71,233,234],{},[74,235,236,239],{},[77,237,238],{},"场景",[77,240,241],{},"推荐",[87,243,244,251,259,267,274,281,288],{},[74,245,246,249],{},[92,247,248],{},"艺术创作",[92,250,100],{},[74,252,253,256],{},[92,254,255],{},"照片级写实",[92,257,258],{},"Flux 2 \u002F Midjourney",[74,260,261,264],{},[92,262,263],{},"图中文字",[92,265,266],{},"Ideogram \u002F GPT Image",[74,268,269,272],{},[92,270,271],{},"Logo 和 UI 素材",[92,273,145],{},[74,275,276,279],{},[92,277,278],{},"游戏角色",[92,280,160],{},[74,282,283,286],{},[92,284,285],{},"商用安全",[92,287,175],{},[74,289,290,293],{},[92,291,292],{},"免费使用",[92,294,295],{},"Microsoft Designer \u002F Ideogram Free",[26,297,298],{"id":298},"推荐工具组合",[227,300,301],{"id":301},"快速创作组合",[30,303,304,307],{},[33,305,306],{},"在线平台出第一版",[33,308,309],{},"满意方向后再补高分辨率或局部重绘",[22,311,312],{},"适合海报、封面、社媒图和灵感探索。",[227,314,315],{"id":315},"本地可控组合",[30,317,318,321,324],{},[33,319,320],{},"模型管理：Civitai \u002F Hugging Face",[33,322,323],{},"工作流：ComfyUI",[33,325,326],{},"素材后处理：FFmpeg \u002F 图像工具 \u002F PS 类软件",[22,328,329],{},"适合需要 LoRA、ControlNet、角色一致性和批量出图的场景。",[26,331,332],{"id":332},"开源模型",[227,334,336],{"id":335},"flux","Flux",[22,338,339],{},"Black Forest Labs 推出的开源文生图模型，目前最强的开源方案：",[68,341,342,351],{},[71,343,344],{},[74,345,346,349],{},[77,347,348],{},"版本",[77,350,82],{},[87,352,353,361,369,377],{},[74,354,355,358],{},[92,356,357],{},"Flux.1 Dev",[92,359,360],{},"开发版，质量高，速度适中",[74,362,363,366],{},[92,364,365],{},"Flux.1 Schnell",[92,367,368],{},"快速版，4 步出图",[74,370,371,374],{},[92,372,373],{},"Flux 2",[92,375,376],{},"最新版，照片级写实",[74,378,379,382],{},[92,380,381],{},"Flux Kontext",[92,383,384],{},"支持图像编辑和角色一致性",[22,386,387],{},"可在 ComfyUI 中使用 Flux 模型：",[389,390,395],"pre",{"className":391,"code":392,"language":393,"meta":394,"style":394},"language-bash shiki shiki-themes github-light github-dark","# 下载 Flux.1 Dev（需 Hugging Face 账号）\n# 放入 ComfyUI\u002Fmodels\u002Fcheckpoints\u002F\n","bash","",[396,397,398,407],"code",{"__ignoreMap":394},[399,400,403],"span",{"class":401,"line":402},"line",1,[399,404,406],{"class":405},"sJ8bj","# 下载 Flux.1 Dev（需 Hugging Face 账号）\n",[399,408,410],{"class":401,"line":409},2,[399,411,412],{"class":405},"# 放入 ComfyUI\u002Fmodels\u002Fcheckpoints\u002F\n",[227,414,416],{"id":415},"stable-diffusion","Stable Diffusion",[68,418,419,427],{},[71,420,421],{},[74,422,423,425],{},[77,424,348],{},[77,426,82],{},[87,428,429,437,445],{},[74,430,431,434],{},[92,432,433],{},"SD 1.5",[92,435,436],{},"经典版，LoRA 生态最丰富",[74,438,439,442],{},[92,440,441],{},"SDXL",[92,443,444],{},"高分辨率，质量提升明显",[74,446,447,450],{},[92,448,449],{},"SD 3.5",[92,451,452],{},"最新版，文字理解增强",[26,454,455],{"id":455},"本地部署",[227,457,459],{"id":458},"stable-diffusion-webui","Stable Diffusion WebUI",[22,461,462],{},"最流行的本地 AI 图像生成界面：",[389,464,466],{"className":391,"code":465,"language":393,"meta":394,"style":394},"# 克隆\ngit clone https:\u002F\u002Fgithub.com\u002FAUTOMATIC1111\u002Fstable-diffusion-webui\ncd stable-diffusion-webui\n\n# 启动（首次会自动下载依赖）\n.\u002Fwebui-user.bat\n",[396,467,468,473,486,496,503,509],{"__ignoreMap":394},[399,469,470],{"class":401,"line":402},[399,471,472],{"class":405},"# 克隆\n",[399,474,475,479,483],{"class":401,"line":409},[399,476,478],{"class":477},"sScJk","git",[399,480,482],{"class":481},"sZZnC"," clone",[399,484,485],{"class":481}," https:\u002F\u002Fgithub.com\u002FAUTOMATIC1111\u002Fstable-diffusion-webui\n",[399,487,489,493],{"class":401,"line":488},3,[399,490,492],{"class":491},"sj4cs","cd",[399,494,495],{"class":481}," stable-diffusion-webui\n",[399,497,499],{"class":401,"line":498},4,[399,500,502],{"emptyLinePlaceholder":501},true,"\n",[399,504,506],{"class":401,"line":505},5,[399,507,508],{"class":405},"# 启动（首次会自动下载依赖）\n",[399,510,512],{"class":401,"line":511},6,[399,513,514],{"class":477},".\u002Fwebui-user.bat\n",[22,516,517,518],{},"访问 ",[396,519,520],{},"http:\u002F\u002Flocalhost:7860",[227,522,524],{"id":523},"comfyui","ComfyUI",[22,526,527],{},"节点式工作流，更灵活的图像生成：",[389,529,531],{"className":391,"code":530,"language":393,"meta":394,"style":394},"git clone https:\u002F\u002Fgithub.com\u002Fcomfyanonymous\u002FComfyUI\ncd ComfyUI\npip install -r requirements.txt\npython main.py\n",[396,532,533,542,549,563],{"__ignoreMap":394},[399,534,535,537,539],{"class":401,"line":402},[399,536,478],{"class":477},[399,538,482],{"class":481},[399,540,541],{"class":481}," https:\u002F\u002Fgithub.com\u002Fcomfyanonymous\u002FComfyUI\n",[399,543,544,546],{"class":401,"line":409},[399,545,492],{"class":491},[399,547,548],{"class":481}," ComfyUI\n",[399,550,551,554,557,560],{"class":401,"line":488},[399,552,553],{"class":477},"pip",[399,555,556],{"class":481}," install",[399,558,559],{"class":491}," -r",[399,561,562],{"class":481}," requirements.txt\n",[399,564,565,568],{"class":401,"line":498},[399,566,567],{"class":477},"python",[399,569,570],{"class":481}," main.py\n",[22,572,517,573],{},[396,574,575],{},"http:\u002F\u002Flocalhost:8188",[227,577,578],{"id":578},"硬件要求",[30,580,581,584,587],{},[33,582,583],{},"显存：最低 4GB（SDXL 建议 8GB+）",[33,585,586],{},"推荐 NVIDIA GPU（CUDA 加速）",[33,588,589],{},"内存：16GB+",[26,591,592],{"id":592},"推荐落地顺序",[22,594,595],{},"建议按下面顺序上手：",[597,598,599,602,605,608,611],"ol",{},[33,600,601],{},"先用在线平台验证提示词和风格方向",[33,603,604],{},"再选一个本地工作流工具",[33,606,607],{},"再下载最少量的基础模型和常用 LoRA",[33,609,610],{},"再补 ControlNet、修脸、放大、局部重绘流程",[33,612,613],{},"最后再做批量生成和风格资产库管理",[22,615,616],{},"这样可以避免一开始就下载一堆模型，结果目录混乱、显存也不够用。",[26,618,619],{"id":619},"模型资源",[227,621,219],{"id":622},"civitai",[22,624,625],{},"AI 图像生成模型社区，包含 Checkpoint、LoRA、VAE 等：",[30,627,628],{},[33,629,630,631],{},"地址：",[95,632,635],{"href":633,"rel":634},"https:\u002F\u002Fcivitai.com\u002Fmodels",[99],"civitai.com\u002Fmodels",[227,637,639],{"id":638},"hugging-face","Hugging Face",[22,641,642],{},"模型托管平台：",[30,644,645],{},[33,646,630,647],{},[95,648,651],{"href":649,"rel":650},"https:\u002F\u002Fhuggingface.co\u002F",[99],"huggingface.co",[227,653,655],{"id":654},"ollama本地大模型","Ollama（本地大模型）",[389,657,661],{"className":658,"code":659,"language":660,"meta":394,"style":394},"language-powershell shiki shiki-themes github-light github-dark","winget install Ollama.Ollama\n\n# 运行图像理解模型\nollama run llava\n","powershell",[396,662,663,669,673,678],{"__ignoreMap":394},[399,664,665],{"class":401,"line":402},[399,666,668],{"class":667},"sVt8B","winget install Ollama.Ollama\n",[399,670,671],{"class":401,"line":409},[399,672,502],{"emptyLinePlaceholder":501},[399,674,675],{"class":401,"line":488},[399,676,677],{"class":405},"# 运行图像理解模型\n",[399,679,680],{"class":401,"line":498},[399,681,682],{"class":667},"ollama run llava\n",[26,684,685],{"id":685},"模型类型",[68,687,688,700],{},[71,689,690],{},[74,691,692,695,697],{},[77,693,694],{},"类型",[77,696,82],{},[77,698,699],{},"大小",[87,701,702,713,724,735,746],{},[74,703,704,707,710],{},[92,705,706],{},"Checkpoint",[92,708,709],{},"基础模型",[92,711,712],{},"2-7 GB",[74,714,715,718,721],{},[92,716,717],{},"LoRA",[92,719,720],{},"微调模型，叠加使用",[92,722,723],{},"10-200 MB",[74,725,726,729,732],{},[92,727,728],{},"VAE",[92,730,731],{},"色彩解码器",[92,733,734],{},"~300 MB",[74,736,737,740,743],{},[92,738,739],{},"Embedding",[92,741,742],{},"文本嵌入",[92,744,745],{},"~100 KB",[74,747,748,751,754],{},[92,749,750],{},"ControlNet",[92,752,753],{},"姿态\u002F边缘控制",[92,755,756],{},"~1.5 GB",[26,758,759],{"id":759},"常见问题",[227,761,762],{"id":762},"出图很糊或细节差",[22,764,765],{},"优先检查：",[30,767,768,771,774,777],{},[33,769,770],{},"基础模型是否适合当前题材",[33,772,773],{},"分辨率是否过低",[33,775,776],{},"采样步数和采样器是否合理",[33,778,779],{},"是否需要高分修复、放大或二次重绘",[227,781,783],{"id":782},"手部文字结构总是崩","手部、文字、结构总是崩",[22,785,786],{},"这是图像生成常见问题。可以优先尝试：",[30,788,789,792,795,798],{},[33,790,791],{},"更强的基础模型",[33,793,794],{},"局部重绘",[33,796,797],{},"ControlNet \u002F 参考图",[33,799,800],{},"专门擅长文字的在线平台",[227,802,803],{"id":803},"本地跑不动",[22,805,806],{},"先不要急着堆插件和大模型。更稳的做法是：",[30,808,809,812,815,818],{},[33,810,811],{},"先跑轻量模型或 Schnell 版本",[33,813,814],{},"降低分辨率",[33,816,817],{},"关闭不必要的附加节点",[33,819,820],{},"确认 CUDA、驱动和显存占用是否正常",[26,822,823],{"id":823},"风险提醒",[30,825,826,829,832,835],{},[33,827,828],{},"模型站内容复杂，下载前先确认来源和许可范围",[33,830,831],{},"商业用途要额外确认平台和模型许可",[33,833,834],{},"LoRA、Embedding、VAE 过多时要做好目录与命名管理",[33,836,837],{},"浏览器在线出图工具方便，但敏感素材和私有项目更适合本地流程",[26,839,840],{"id":840},"延伸阅读",[30,842,843,849,855],{},[33,844,845],{},[95,846,848],{"href":847},"\u002Fdocs\u002Fai-tools","AI 工具合集",[33,850,851],{},[95,852,854],{"href":853},"\u002Fdocs\u002Fimage-video-processing","图像与视频处理",[33,856,857],{},[95,858,860],{"href":859},"\u002Fdocs\u002Fdesign-resources","设计资源库",[26,862,863],{"id":863},"提示词技巧",[227,865,866],{"id":866},"正面提示词结构",[389,868,873],{"className":869,"code":871,"language":872},[870],"language-text","主题描述, 画面质量, 风格, 光照, 细节\n\n例：\n1girl, white dress, standing in flower field,\nmasterpiece, best quality, ultra detailed,\nsoft lighting, bokeh, depth of field\n","text",[396,874,871],{"__ignoreMap":394},[227,876,877],{"id":877},"负面提示词",[389,879,882],{"className":880,"code":881,"language":872},[870],"lowres, bad anatomy, bad hands, text, error,\nmissing fingers, extra digit, fewer digits,\ncropped, worst quality, low quality, blurry\n",[396,883,881],{"__ignoreMap":394},[227,885,886],{"id":886},"常用参数",[68,888,889,901],{},[71,890,891],{},[74,892,893,896,899],{},[77,894,895],{},"参数",[77,897,898],{},"推荐值",[77,900,82],{},[87,902,903,914,925,936],{},[74,904,905,908,911],{},[92,906,907],{},"Steps",[92,909,910],{},"20-30",[92,912,913],{},"采样步数",[74,915,916,919,922],{},[92,917,918],{},"CFG Scale",[92,920,921],{},"7-9",[92,923,924],{},"提示词相关度",[74,926,927,930,933],{},[92,928,929],{},"Sampler",[92,931,932],{},"DPM++ 2M Karras",[92,934,935],{},"采样器",[74,937,938,941,944],{},[92,939,940],{},"分辨率",[92,942,943],{},"512x512 \u002F 1024x1024",[92,945,946],{},"根据模型选择",[26,948,949],{"id":949},"参考链接",[30,951,952,959,965,971,977,983,989,997],{},[33,953,954,958],{},[95,955,459],{"href":956,"rel":957},"https:\u002F\u002Fgithub.com\u002FAUTOMATIC1111\u002Fstable-diffusion-webui",[99]," — GitHub",[33,960,961,958],{},[95,962,524],{"href":963,"rel":964},"https:\u002F\u002Fgithub.com\u002Fcomfyanonymous\u002FComfyUI",[99],[33,966,967,970],{},[95,968,219],{"href":217,"rel":969},[99]," — 模型社区",[33,972,973,976],{},[95,974,100],{"href":97,"rel":975},[99]," — 官网",[33,978,979,982],{},[95,980,130],{"href":128,"rel":981},[99]," — 文字生成最强",[33,984,985,988],{},[95,986,145],{"href":143,"rel":987},[99]," — 设计师向工具",[33,990,991,996],{},[95,992,995],{"href":993,"rel":994},"https:\u002F\u002Fhuggingface.co\u002Fblack-forest-labs",[99],"Flux Models"," — Hugging Face",[33,998,999,1004],{},[95,1000,1003],{"href":1001,"rel":1002},"https:\u002F\u002Fgithub.com\u002Fkohya-ss\u002Fsd-scripts",[99],"LoRA 训练教程"," — Kohya 训练脚本",[1006,1007,1008],"style",{},"html pre.shiki code .sJ8bj, html code.shiki .sJ8bj{--shiki-default:#6A737D;--shiki-dark:#6A737D}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html pre.shiki code .sScJk, html code.shiki .sScJk{--shiki-default:#6F42C1;--shiki-dark:#B392F0}html pre.shiki code .sZZnC, html code.shiki .sZZnC{--shiki-default:#032F62;--shiki-dark:#9ECBFF}html pre.shiki code .sj4cs, html code.shiki .sj4cs{--shiki-default:#005CC5;--shiki-dark:#79B8FF}html pre.shiki code .sVt8B, html code.shiki .sVt8B{--shiki-default:#24292E;--shiki-dark:#E1E4E8}",{"title":394,"searchDepth":409,"depth":409,"links":1010},[1011,1012,1015,1019,1023,1028,1029,1034,1035,1040,1041,1042,1047],{"id":28,"depth":409,"text":28},{"id":66,"depth":409,"text":66,"children":1013},[1014],{"id":229,"depth":488,"text":229},{"id":298,"depth":409,"text":298,"children":1016},[1017,1018],{"id":301,"depth":488,"text":301},{"id":315,"depth":488,"text":315},{"id":332,"depth":409,"text":332,"children":1020},[1021,1022],{"id":335,"depth":488,"text":336},{"id":415,"depth":488,"text":416},{"id":455,"depth":409,"text":455,"children":1024},[1025,1026,1027],{"id":458,"depth":488,"text":459},{"id":523,"depth":488,"text":524},{"id":578,"depth":488,"text":578},{"id":592,"depth":409,"text":592},{"id":619,"depth":409,"text":619,"children":1030},[1031,1032,1033],{"id":622,"depth":488,"text":219},{"id":638,"depth":488,"text":639},{"id":654,"depth":488,"text":655},{"id":685,"depth":409,"text":685},{"id":759,"depth":409,"text":759,"children":1036},[1037,1038,1039],{"id":762,"depth":488,"text":762},{"id":782,"depth":488,"text":783},{"id":803,"depth":488,"text":803},{"id":823,"depth":409,"text":823},{"id":840,"depth":409,"text":840},{"id":863,"depth":409,"text":863,"children":1043},[1044,1045,1046],{"id":866,"depth":488,"text":866},{"id":877,"depth":488,"text":877},{"id":886,"depth":488,"text":886},{"id":949,"depth":409,"text":949},{"path":1049,"title":1050,"description":1051,"docType":8,"resourceKind":9,"categoryId":10,"categoryLabel":11,"updatedAt":12,"publishedAt":12,"icon":13},"\u002Fdocs\u002Fprompt-engineering","AI 提示词工程","ChatGPT\u002FClaude 提示词技巧、角色设定、思维链与常用模板",{"path":1053,"title":1054,"description":1055,"docType":8,"resourceKind":9,"categoryId":1056,"categoryLabel":1057,"updatedAt":12,"publishedAt":12,"icon":1058},"\u002Fdocs\u002Fapi-testing","API 测试与调试","curl、HTTPie、Postman、Bruno 等 API 测试工具与常用技巧","dev-environment","开发环境","i-carbon-code",[1060,1061,1062],"希望把零散经验整理成长期可复用工作流的人","正在使用 AI 工具、Agent 或自动化工作流的人","希望阅读时顺手建立自己的操作清单或收藏体系的人",[1064,1065,1066],"先浏览标题、摘要和目录，带着问题阅读会更高效","确认模型供应商、API Key、CLI 工具链与本地资源是否已准备好","如果页面里提到相关文档，尽量一起打开对照，效果通常更完整",[1068,1073,1078,1082],{"path":1069,"title":1070,"description":1071,"docType":8,"resourceKind":9,"categoryId":10,"categoryLabel":11,"updatedAt":1072,"publishedAt":1072,"icon":13},"\u002Fdocs\u002Fskills-guide","AI Agent Skills 指南","理解 skills 的作用、目录结构、编写方式，以及它与 MCP 的关系","2026-03-08",{"path":1074,"title":1075,"description":1076,"docType":8,"resourceKind":9,"categoryId":10,"categoryLabel":11,"updatedAt":1077,"publishedAt":1077,"icon":13},"\u002Fdocs\u002Fai-local-models","本地 AI 模型部署","Ollama、LM Studio、vLLM 本地大模型运行与 API 调用","2026-02-28",{"path":1079,"title":1080,"description":1081,"docType":8,"resourceKind":9,"categoryId":10,"categoryLabel":11,"updatedAt":1077,"publishedAt":1077,"icon":13},"\u002Fdocs\u002Flocal-llm-deployment","本地 LLM 部署指南","使用 Ollama、vLLM、LM Studio 在本地运行大语言模型",{"path":1083,"title":1084,"description":1085,"docType":8,"resourceKind":9,"categoryId":10,"categoryLabel":11,"updatedAt":1077,"publishedAt":1077,"icon":13},"\u002Fdocs\u002Fai-coding-rules","AI 编程助手规则配置","Cursor Rules、Claude Projects、Kiro Steering 等 AI 编程助手的规则与上下文配置",1776215711221]