云端 AI Agent 架构图

{"type":"企业级云端 AI 架构信息图","style":"简洁的企业矢量图，白色背景，蓝色与青色调色板，圆角矩形，细蓝色轮廓，柔和阴影，以中文为主的双语标签并配有较小的英文副标题，对称布局，演示文稿级质量","canvas":{"aspect_ratio":"4:3","resolution":"1024x768"},"title_block":{"headline":"{argument name=\"headline text\" default=\"云端 AI Agent 架构图\"}","subtitle":"面向云端任务执行、代码协作与大规模调度的 Cloud AI Agent Architecture","position":"top center"},"layout":{"center":{"type":"堆叠式架构层","count":7,"numbered":true,"layers":[{"index":1,"title":"接入与体验层","english":"access and experience layer","count":5,"items":["Web Console","IDE / CLI","API / SDK","消息通知 Slack / Email","任务看板 组织与项目空间"]},{"index":2,"title":"云端控制平面","english":"cloud control plane","count":6,"items":["API Gateway","Auth / Tenant","Session Manager","Project / Workspace Manager","Task Queue","Rate Limit / Billing"]},{"index":3,"title":"Agent 编排层","english":"agent orchestration layer","count":8,"items":["Planner / 任务规划器","Reasoner / 推理引擎","Context Manager / 上下文管理","Memory / 记忆系统","Prompt Builder","Workflow Orchestrator / 工作流编排","Policy Engine / 策略引擎","Multi-Agent Coordinator / 多 Agent 协同"]},{"index":4,"title":"执行运行时层","english":"execution runtime layer","count":8,"items":["Sandbox Pool / 容器沙箱","Browser / HTTP 工具","Code Runner / Terminal","Git Workspace","Tool Adapters / MCP","Scheduler / Worker","Test Runner","Artifact Executor / Patch & Diff"]},{"index":5,"title":"模型与智能服务层","english":"model and intelligence services layer","count":6,"items":["Model Router","Frontier LLM API","Embedding / Reranker","OCR / STT / TTS（可选）","Evaluator / Reflection","第三方 Agent / SaaS"]},{"index":6,"title":"数据与状态层","english":"data and state layer","count":8,"items":["Relational DB","Vector DB","Redis Cache","Session State","Artifact Store","对象存储 Object Storage","配置中心 Config","日志与审计 Logs"]},{"index":7,"title":"云基础设施层","english":"cloud infrastructure layer","count":8,"items":["Kubernetes / 容器编排","Service Mesh","Event Bus","Secrets / KMS","VPC / Network","Autoscaling","CI/CD","多区域容灾 DR"]}]},"left_sidebar":{"title":"可观测性 / SRE","position":"left","count":6,"items":["监控 Monitoring","Tracing","Metrics","日志聚合 Logs","成本与配额 Cost / Quota","告警 Alerting"]},"right_sidebar":{"title":"安全治理","position":"right","count":6,"items":["SSO / RBAC","密钥与 Secrets","策略护栏 Guardrails","人工审批 Human Approval","审计合规 Audit","数据隔离 Isolation"]},"connectors":"连接两侧边栏至每个中心层的双向蓝色箭头，以及各架构层之间的垂直箭头"},"visual_details":{"icons":"每个方框内配有简洁的蓝色线条图标，每个模块一个图标","boxes":"带有浅蓝或浅青色背景的圆角白色卡片","section_colors":{"layers_1_2_5_7":"浅蓝色","layer_3":"浅紫色","layer_4":"浅青色","layer_6":"浅绿色","sidebars":"深宝蓝色标题栏，白色文字"},"typography":"加粗深海军蓝中文标题，较小的灰蓝色副标题，中等字重中文分区标签，紧凑的双语项目标签"},"footer":{"position":"bottom center","text":"{argument name=\"footer text\" default=\"特性：云原生 • 多租户 • 弹性扩缩容 • 安全合规 • 人机协同 • 任务可追踪 • 支持多模型与多 Agent 协同\"}"},"quality":"清晰的信息图，布局平衡，高可读性，适用于 Keynote、文档及产品营销的专业企业架构图"}

• 模型gpt-image-2
• 用途social-media-post
• 风格minimal, editorial
• 画幅Flexible

• 用途social-media-post
• 风格minimal, editorial
• 推荐规格1:1, 4:5, 16:9, or 9:16
• 模型GPT Image 2

把这个 云端 AI Agent 架构图 提示词当作起点。替换主体、产品细节、场景、光线、品牌颜色和输出画幅，同时保留 social-media-post 的用途和 minimal, editorial 的视觉方向。

• 主体替换成你的产品、人物、场景或界面对象。
• 用途保留或改写为 social media post、广告图、封面、海报或电商主图。
• 画幅按投放位置选择 1:1、4:5、16:9 或 9:16。
• 品牌线索补充颜色、材质、道具、文案安全区和最终使用平台。

先按目标平台选择画幅，再根据用途选择 1K、2K 或 4K。产品图和海报通常适合更高清晰度。

• 检查主体、光线、构图和画幅都足够明确。
• 检查如果图片要承载标题或广告文案，已经预留文字安全区。
• 检查生成前确认分辨率、额度成本和最终使用位置。

云端 AI Agent 架构图 的重点是先固定图片用途，再用主体、构图、光线、风格和画幅控制生成结果。它适合 social media post，并保留 minimal, editorial 的视觉方向。

• 能直接用吗？可以。先替换主体、品牌线索和画幅，再进入 Image2Studio 生成。
• 适合什么比例？按平台选择 1:1、4:5、16:9 或 9:16。