[{"data":1,"prerenderedAt":642},["ShallowReactive",2],{"navigation":3,"\u002Fmodels\u002Ftool-use":312,"\u002Fmodels\u002Ftool-use-surround":637},[4,21,54,95,153,169,188,251,280,292],{"title":5,"path":6,"stem":7,"children":8,"icon":20},"快速开始","\u002Fstart","1.start\u002F0.index",[9,11,15],{"title":5,"path":6,"stem":7,"icon":10},"i-lucide-house",{"title":12,"path":13,"stem":14,"icon":10},"使用 TokenResearch Docs","\u002Fstart\u002Felysia-code","1.start\u002F1.elysia-code",{"title":16,"path":17,"stem":18,"icon":19},"安装Node.js 和 Git","\u002Fstart\u002Fnode-git","1.start\u002F2.node-git","i-simple-icons-javascript",false,{"title":22,"path":23,"stem":24,"children":25,"icon":20},"基础教程","\u002Ftutorials","2.tutorials\u002F1.index",[26,29,34,39,44,49],{"title":27,"path":23,"stem":24,"icon":28},"教程路线图","i-lucide-route",{"title":30,"path":31,"stem":32,"icon":33},"环境准备","\u002Ftutorials\u002Fenvironment","2.tutorials\u002F2.environment","i-lucide-settings-2",{"title":35,"path":36,"stem":37,"icon":38},"Codex 首次运行","\u002Ftutorials\u002Fcodex-first-run","2.tutorials\u002F3.codex-first-run","i-lucide-terminal",{"title":40,"path":41,"stem":42,"icon":43},"Agent 规则","\u002Ftutorials\u002Fagent-rules","2.tutorials\u002F4.agent-rules","i-lucide-scroll-text",{"title":45,"path":46,"stem":47,"icon":48},"视觉验收","\u002Ftutorials\u002Fvisual-review","2.tutorials\u002F5.visual-review","i-lucide-monitor-check",{"title":50,"path":51,"stem":52,"icon":53},"质量门禁","\u002Ftutorials\u002Fquality-gates","2.tutorials\u002F6.quality-gates","i-lucide-shield-check",{"title":55,"path":56,"stem":57,"children":58,"icon":20},"Codex 配置文档","\u002Fcodex","3.codex\u002F0.index",[59,61,66,71,76,80,85,90],{"title":55,"path":56,"stem":57,"icon":60},"i-lucide-bot",{"title":62,"path":63,"stem":64,"icon":65},"安装Codex CLI","\u002Fcodex\u002Finstall-cli","3.codex\u002F1.install-cli","i-simple-icons-openai",{"title":67,"path":68,"stem":69,"icon":70},"安装 IDE 插件","\u002Fcodex\u002Finstall-ide","3.codex\u002F2.install-ide","i-lucide-puzzle",{"title":72,"path":73,"stem":74,"icon":75},"Codex App桌面端","\u002Fcodex\u002Fapp-desktop","3.codex\u002F3.app-desktop","i-lucide-flask-conical",{"title":77,"path":78,"stem":79,"icon":60},"Codex配置\u002F使用","\u002Fcodex\u002Fmodel","3.codex\u002F5.model",{"title":81,"path":82,"stem":83,"icon":84},"其他错误","\u002Fcodex\u002Fother-errors","3.codex\u002F6.other-errors","i-lucide-info",{"title":86,"path":87,"stem":88,"icon":89},"网络与代理","\u002Fcodex\u002Fnetwork-proxy","3.codex\u002F7.network-proxy","i-lucide-globe",{"title":91,"path":92,"stem":93,"icon":94},"Codex模型测试","\u002Fcodex\u002Fmodel-test","3.codex\u002F8.model-test","i-lucide-boxes",{"title":96,"path":97,"stem":98,"children":99,"icon":20},"模型中心","\u002Fmodels","4.models\u002F01.index",[100,102,106,111,116,121,125,129,134,139,144,149],{"title":96,"path":97,"stem":98,"icon":101},"i-lucide-store",{"title":103,"path":104,"stem":105,"icon":28},"任务路由建议","\u002Fmodels\u002Ftask-routing","4.models\u002F02.task-routing",{"title":107,"path":108,"stem":109,"icon":110},"推理等级评测","\u002Fmodels\u002Freasoning-levels","4.models\u002F03.reasoning-levels","i-lucide-brain",{"title":112,"path":113,"stem":114,"icon":115},"上下文与压缩","\u002Fmodels\u002Fcontext-compression","4.models\u002F04.context-compression","i-lucide-file-stack",{"title":117,"path":118,"stem":119,"icon":120},"评测记录模板","\u002Fmodels\u002Feval-template","4.models\u002F05.eval-template","i-lucide-clipboard-list",{"title":122,"path":123,"stem":124,"icon":48},"前端视觉评测","\u002Fmodels\u002Ffrontend-visual","4.models\u002F06.frontend-visual",{"title":126,"path":127,"stem":128,"icon":43},"研究写作评测","\u002Fmodels\u002Fresearch-writing","4.models\u002F07.research-writing",{"title":130,"path":131,"stem":132,"icon":133},"运维排查评测","\u002Fmodels\u002Fops-debug","4.models\u002F08.ops-debug","i-lucide-server-cog",{"title":135,"path":136,"stem":137,"icon":138},"成本与速度评测","\u002Fmodels\u002Fcost-speed","4.models\u002F09.cost-speed","i-lucide-timer",{"title":140,"path":141,"stem":142,"icon":143},"工具调用评测","\u002Fmodels\u002Ftool-use","4.models\u002F10.tool-use","i-lucide-terminal-square",{"title":145,"path":146,"stem":147,"icon":148},"样例评测记录","\u002Fmodels\u002Fsample-records","4.models\u002F11.sample-records","i-lucide-clipboard-check",{"title":150,"path":151,"stem":152,"icon":101},"模型广场","\u002Fmodels\u002Fmarket","4.models\u002F12.market",{"title":154,"path":155,"stem":156,"children":157,"icon":20},"OpenCode 配置文档","\u002Fopencode","5.opencode\u002F0.index",[158,160,165],{"title":154,"path":155,"stem":156,"icon":159},"i-lucide-code-xml",{"title":161,"path":162,"stem":163,"icon":164},"安装OpenCode","\u002Fopencode\u002Finstall-opencode","5.opencode\u002F1.install-opencode","i-lucide-code",{"title":166,"path":167,"stem":168,"icon":164},"配置模型","\u002Fopencode\u002Fmodel","5.opencode\u002F2.model",{"title":170,"path":171,"stem":172,"children":173,"icon":20},"Claude Code 配置文档","\u002Fclaude-code","6.claude-code\u002F0.index",[174,176,180,184],{"title":170,"path":171,"stem":172,"icon":175},"i-lucide-sparkles",{"title":177,"path":178,"stem":179,"icon":175},"安装Claude CLI","\u002Fclaude-code\u002Finstall-cli","6.claude-code\u002F1.install-cli",{"title":181,"path":182,"stem":183,"icon":70},"插件配置","\u002Fclaude-code\u002Fplugin","6.claude-code\u002F2.plugin",{"title":185,"path":186,"stem":187,"icon":94},"模型配置","\u002Fclaude-code\u002Fmodel","6.claude-code\u002F3.model",{"title":189,"path":190,"stem":191,"children":192,"icon":20},"Skill Gallery","\u002Fgallery","7.gallery\u002F01.index",[193,195,200,205,210,215,219,224,229,234,238,242,247],{"title":189,"path":190,"stem":191,"icon":194},"i-lucide-gallery-horizontal-end",{"title":196,"path":197,"stem":198,"icon":199},"Image Generation","\u002Fgallery\u002Fimagegen","7.gallery\u002F02.imagegen","i-lucide-image",{"title":201,"path":202,"stem":203,"icon":204},"Paper Analyze","\u002Fgallery\u002Fpaper-analyze","7.gallery\u002F03.paper-analyze","i-lucide-file-search",{"title":206,"path":207,"stem":208,"icon":209},"Presentation Builder","\u002Fgallery\u002Fpresentation","7.gallery\u002F04.presentation","i-lucide-presentation",{"title":211,"path":212,"stem":213,"icon":214},"Spreadsheet Analyst","\u002Fgallery\u002Fspreadsheet","7.gallery\u002F05.spreadsheet","i-lucide-table-properties",{"title":216,"path":217,"stem":218,"icon":48},"Browser Visual QA","\u002Fgallery\u002Fbrowser-visual-qa","7.gallery\u002F06.browser-visual-qa",{"title":220,"path":221,"stem":222,"icon":223},"Zotero Research","\u002Fgallery\u002Fzotero-research","7.gallery\u002F07.zotero-research","i-lucide-library",{"title":225,"path":226,"stem":227,"icon":228},"Research Loop","\u002Fgallery\u002Fresearch-loop","7.gallery\u002F08.research-loop","i-lucide-orbit",{"title":230,"path":231,"stem":232,"icon":233},"PDF and Docs","\u002Fgallery\u002Fpdf-docs","7.gallery\u002F09.pdf-docs","i-lucide-file-type",{"title":235,"path":236,"stem":237,"icon":133},"Ops Debug","\u002Fgallery\u002Fops-debug","7.gallery\u002F10.ops-debug",{"title":239,"path":240,"stem":241,"icon":43},"LaTeX Paper","\u002Fgallery\u002Flatex-paper","7.gallery\u002F11.latex-paper",{"title":243,"path":244,"stem":245,"icon":246},"Diagram Canvas","\u002Fgallery\u002Fdiagram-canvas","7.gallery\u002F12.diagram-canvas","i-lucide-git-compare-arrows",{"title":248,"path":249,"stem":250,"icon":175},"Hatch Pet","\u002Fgallery\u002Fhatch-pet","7.gallery\u002F13.hatch-pet",{"title":252,"path":253,"stem":254,"children":255,"icon":20},"场景路线","\u002Fworkflows","8.workflows\u002F01.index",[256,258,263,268,272,276],{"title":252,"path":253,"stem":254,"icon":257},"i-lucide-map",{"title":259,"path":260,"stem":261,"icon":262},"论文到汇报","\u002Fworkflows\u002Fpaper-to-slides","8.workflows\u002F02.paper-to-slides","i-lucide-file-chart-column",{"title":264,"path":265,"stem":266,"icon":267},"视觉复刻路线","\u002Fworkflows\u002Fvisual-replica","8.workflows\u002F03.visual-replica","i-lucide-monitor-dot",{"title":269,"path":270,"stem":271,"icon":28},"模型路由记录","\u002Fworkflows\u002Fmodel-routing","8.workflows\u002F04.model-routing",{"title":273,"path":274,"stem":275,"icon":194},"Skill 沉淀路线","\u002Fworkflows\u002Fskill-to-gallery","8.workflows\u002F05.skill-to-gallery",{"title":277,"path":278,"stem":279,"icon":133},"运维现场排查","\u002Fworkflows\u002Fops-live-debug","8.workflows\u002F06.ops-live-debug",{"title":281,"path":282,"stem":283,"children":284,"icon":20},"常见问题","\u002Ffaq","91.faq\u002F0.index",[285,288],{"title":286,"path":282,"stem":283,"icon":287},"FAQ","i-lucide-circle-help",{"title":289,"path":290,"stem":291,"icon":84},"通用问题","\u002Ffaq\u002Fcommon","91.faq\u002F1.common",{"title":293,"path":294,"stem":295,"children":296,"icon":20},"Skills 与 MCP","\u002Fskills-mcp","92.skills-mcp\u002F0.index",[297,299,304,308],{"title":293,"path":294,"stem":295,"icon":298},"i-lucide-plug",{"title":300,"path":301,"stem":302,"icon":303},"Skills 推荐","\u002Fskills-mcp\u002Fskills","92.skills-mcp\u002F1.skills","i-lucide-link",{"title":305,"path":306,"stem":307,"icon":298},"MCP 使用边界","\u002Fskills-mcp\u002Fmcp-overview","92.skills-mcp\u002F2.mcp-overview",{"title":309,"path":310,"stem":311,"icon":303},"chrome-devtools-mcp","\u002Fskills-mcp\u002Fchrome-devtools-mcp","92.skills-mcp\u002F3.chrome-devtools-mcp",{"id":313,"title":140,"body":314,"description":630,"extension":631,"links":632,"meta":633,"navigation":634,"path":141,"seo":635,"stem":142,"__hash__":636},"docs\u002F4.models\u002F10.tool-use.md",{"type":315,"value":316,"toc":618},"minimark",[317,321,325,328,393,396,406,409,412,428,431,477,480,532,535,612,615],[318,319,320],"h2",{"id":320},"评测目标",[322,323,324],"p",{},"AI 编程工具的关键能力不是单纯生成代码，而是能否在真实项目里正确使用工具：读对文件、改对边界、运行检查、根据错误继续修复。",[318,326,327],{"id":327},"任务集",[329,330,331,344],"table",{},[332,333,334],"thead",{},[335,336,337,341],"tr",{},[338,339,340],"th",{},"任务",[338,342,343],{},"观察点",[345,346,347,361,369,377,385],"tbody",{},[335,348,349,353],{},[350,351,352],"td",{},"文件定位",[350,354,355,356,360],{},"是否先用 ",[357,358,359],"code",{},"rg"," 找入口",[335,362,363,366],{},[350,364,365],{},"精准编辑",[350,367,368],{},"是否只改相关文件",[335,370,371,374],{},[350,372,373],{},"命令执行",[350,375,376],{},"是否用项目已有脚本",[335,378,379,382],{},[350,380,381],{},"错误恢复",[350,383,384],{},"是否能根据 lint\u002Ftypecheck\u002Fbuild 修复",[335,386,387,390],{},[350,388,389],{},"浏览器检查",[350,391,392],{},"是否能保存截图并判断视觉问题",[318,394,395],{"id":395},"最小评测任务",[397,398,404],"pre",{"className":399,"code":401,"language":402,"meta":403},[400],"language-text","在当前 Nuxt 文档站新增一个 Markdown 页面，把它接入导航。\n完成后运行 lint、typecheck、build、route check，并用浏览器截一张移动端截图。\n","text","",[357,405,401],{"__ignoreMap":403},[322,407,408],{},"这个任务能同时覆盖文件搜索、内容编辑、命令验证、浏览器工具和错误恢复。",[318,410,411],{"id":411},"失败信号",[413,414,415,419,422,425],"ul",{},[416,417,418],"li",{},"在不知道项目结构时直接大改。",[416,420,421],{},"忽略用户指定的文件或路径。",[416,423,424],{},"测试失败后只解释不修。",[416,426,427],{},"把截图检查当成可选项。",[318,429,430],{"id":430},"评分方式",[329,432,433,443],{},[332,434,435],{},[335,436,437,440],{},[338,438,439],{},"分数",[338,441,442],{},"标准",[345,444,445,453,461,469],{},[335,446,447,450],{},[350,448,449],{},"0",[350,451,452],{},"没读项目结构，直接输出泛泛建议",[335,454,455,458],{},[350,456,457],{},"1",[350,459,460],{},"能修改文件，但不跑验证",[335,462,463,466],{},[350,464,465],{},"2",[350,467,468],{},"能跑部分命令，失败后需要人工接管",[335,470,471,474],{},[350,472,473],{},"3",[350,475,476],{},"能完整修复并留下命令、路由和截图证据",[318,478,479],{"id":479},"细分评分",[329,481,482,491],{},[332,483,484],{},[335,485,486,489],{},[338,487,488],{},"维度",[338,490,430],{},[345,492,493,500,508,516,524],{},[335,494,495,497],{},[350,496,352],{},[350,498,499],{},"0 分：猜文件；1 分：读到相关文件；2 分：能找入口和调用关系；3 分：能保护用户指定边界",[335,501,502,505],{},[350,503,504],{},"编辑质量",[350,506,507],{},"0 分：大范围无关改动；1 分：能改目标文件；2 分：改动小且符合项目风格；3 分：兼顾代码、内容和后续维护",[335,509,510,513],{},[350,511,512],{},"命令验证",[350,514,515],{},"0 分：不运行；1 分：只跑一个命令；2 分：跑主要命令但不处理失败；3 分：失败后定位并重跑到通过",[335,517,518,521],{},[350,519,520],{},"浏览器使用",[350,522,523],{},"0 分：不打开页面；1 分：只看一张截图；2 分：多视口截图；3 分：结合 Console、截图和 DOM 证据",[335,525,526,529],{},[350,527,528],{},"交付说明",[350,530,531],{},"0 分：只有结论；1 分：列了文件；2 分：列文件和命令；3 分：列命令、截图、风险和下一步",[318,533,534],{"id":534},"工具边界",[329,536,537,550],{},[332,538,539],{},[335,540,541,544,547],{},[338,542,543],{},"工具",[338,545,546],{},"用途",[338,548,549],{},"常见误用",[345,551,552,564,577,590,601],{},[335,553,554,558,561],{},[350,555,556],{},[357,557,359],{},[350,559,560],{},"快速定位文件和文本",[350,562,563],{},"用全盘慢扫代替项目内搜索",[335,565,566,571,574],{},[350,567,568],{},[357,569,570],{},"apply_patch",[350,572,573],{},"小范围手工编辑",[350,575,576],{},"通过脚本重写不该碰的文件",[335,578,579,584,587],{},[350,580,581],{},[357,582,583],{},"pnpm lint\u002Ftypecheck\u002Fbuild",[350,585,586],{},"验证工程状态",[350,588,589],{},"失败后只解释不修",[335,591,592,595,598],{},[350,593,594],{},"Playwright \u002F 浏览器",[350,596,597],{},"检查真实页面",[350,599,600],{},"只截图首屏，不看移动端和 Console",[335,602,603,606,609],{},[350,604,605],{},"SSH \u002F CLI",[350,607,608],{},"检查 live state",[350,610,611],{},"不读配置和日志就改服务",[318,613,614],{"id":614},"验收方式",[322,616,617],{},"模型评测记录里必须保存命令结果、关键截图路径和最终路由 200 检查。没有这些证据，不应认为评测完成。",{"title":403,"searchDepth":619,"depth":620,"links":621},1,2,[622,623,624,625,626,627,628,629],{"id":320,"depth":620,"text":320},{"id":327,"depth":620,"text":327},{"id":395,"depth":620,"text":395},{"id":411,"depth":620,"text":411},{"id":430,"depth":620,"text":430},{"id":479,"depth":620,"text":479},{"id":534,"depth":620,"text":534},{"id":614,"depth":620,"text":614},"评估模型是否能正确读文件、改文件、跑命令和使用浏览器。","md",null,{},{"icon":143},{"title":140,"description":630},"MUQxM7kClG78GflXLnJXQuEUzNOFySZXVqYIfKbk-RQ",[638,640],{"title":135,"path":136,"stem":137,"description":639,"icon":138,"children":-1},"比较不同模型路由在速度、成本和失败重试上的真实开销。",{"title":145,"path":146,"stem":147,"description":641,"icon":148,"children":-1},"用填好的记录展示前端、研究写作和运维排查如何评估模型。",1783590616542]