[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-1928":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":15,"forks30d":15,"starsTrendScore":17,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":23,"hasPages":21,"topics":24,"createdAt":9,"pushedAt":9,"updatedAt":25,"readmeContent":26,"aiSummary":27,"trendingCount":15,"starSnapshotCount":15,"syncStatus":28,"lastSyncTime":29,"discoverSource":30},1928,"engineering-figure-agent","heyu-233\u002Fengineering-figure-agent","heyu-233","Engineering Figure Banana is a Codex skill for generating publication-style engineering and computer science figures, with support for Chinese academic papers, bilingual labeling, and editable-figure handoff workflows.",null,"Python",225,9,145,1,0,7,21,73,3,"MIT License",false,"main",true,[],"2026-06-12 02:00:34","# Engineering Figure Agent\n\n\u003Cdiv align=\"center\">\n\n![License](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-2563eb)\n![Codex Skill](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FCodex-skill-111827)\n![Figure Modes](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fmodes-image%20%7C%20plot-f59e0b)\n![Backends](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fbackends-Gemini%20%7C%20OpenAI%20%7C%20local%20plots-16a34a)\n![Focus](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ffocus-engineering%20papers-7c3aed)\n\n[![Try Engineering Figure Agent on Socialistic](https:\u002F\u002Fsocialistic.ai\u002Fapi\u002Fembed\u002Fengineering-figure-agent-407b80)](https:\u002F\u002Fsocialistic.ai\u002Fzh\u002Fskill\u002Fengineering-figure-agent-407b80?utm_source=github&utm_medium=readme&utm_campaign=20260519-student-research-tools&utm_content=badge)\n\n**Agent-native figure production for engineering and CS papers.**\n\n把工程论文里的系统架构图、算法流程图、实验曲线和多面板图，拆成可控的生成流程：概念图走 image mode，精确数值图走 plot mode。\n\n[中文说明](.\u002FREADME.zh-CN.md) | [English Guide](.\u002FREADME.en.md) | [Example Gallery](.\u002Fdocs\u002Fexamples\u002FREADME.md) | [Showcase](.\u002Fdocs\u002Fshowcase.md)\n\n\u003C\u002Fdiv>\n\n## Preview\n\n| System \u002F Architecture | Cooperative Perception | Safety Taxonomy |\n|---|---|---|\n| ![Federated open-vocabulary driving figure](docs\u002Fexamples\u002Ffederated-open-vocab-driving-2k-1.png) | ![Cooperative object tracking figure](docs\u002Fexamples\u002Fcooperative-object-tracking-2k-1.png) | ![Multi-agent safety overview figure](docs\u002Fexamples\u002Fmulti-agent-safety-overview-2k-1.png) |\n\n| Dense Systems Diagram | Deployment Scenario |\n|---|---|\n| ![Linux kernel system diagram](docs\u002Fexamples\u002Flinux-kernel-system-1.jpg) | ![Health monitoring and early warning deployment scenarios](docs\u002Fexamples\u002Fhealth-monitoring-early-warning-reference.jpg) |\n\n| Exact Local Plot |\n|---|\n| ![Benchmark plot](docs\u002Fexamples\u002Fbenchmark-plot.png) |\n\n## Why It Exists\n\nMost figure tools treat every paper figure as the same image prompt problem. Engineering papers do not work that way.\n\n| Figure need | Better path |\n|---|---|\n| System architecture, pipeline, schematic, graphical abstract | `image mode` with an engineering-aware prompt template |\n| Benchmark curves, ablation bars, heatmaps, scatter plots | `plot mode` with local plotting and exact values |\n| Mixed conceptual + quantitative figure | Render numeric panels locally, then compose or describe the conceptual panels |\n| Reference-inspired redraw | Use image mode for structure and style exploration, then manually verify labels and layout |\n\nEngineering Figure Agent is intentionally lighter than a full paper-upload platform. It is built for researchers who already know what claim a figure should support and want a cleaner production path inside an agent workflow.\n\n## Quick Start\n\nInstall and run the setup check:\n\n```powershell\n& \"$HOME\u002F.codex\u002Fskills\u002Fengineering-figure-agent\u002Fscripts\u002Finstall_and_test.ps1\" -RunSetupCheck\n```\n\nOpen the wizard:\n\n```powershell\n& \"$HOME\u002F.codex\u002Fskills\u002Fengineering-figure-agent\u002Fscripts\u002Fwizard.ps1\"\n```\n\nOr generate a conceptual figure directly:\n\n```powershell\npython \"$HOME\u002F.codex\u002Fskills\u002Fengineering-figure-agent\u002Fscripts\u002Fgenerate_image.py\" `\n  --figure-template system-architecture `\n  --lang en `\n  \"A retrieval-augmented generation system with OCR, chunking, embedding, vector search, reranking, and answer synthesis.\"\n```\n\nFor installation details, use the language-specific guides:\n\n- [README.zh-CN.md](.\u002FREADME.zh-CN.md): Windows 最短安装路径、验证方法、中文示例\n- [README.en.md](.\u002FREADME.en.md): setup, verification, examples, positioning\n\n## Two Modes\n\n### Image Mode\n\nUse `image mode` when visual structure matters more than exact numeric geometry.\n\nBest for:\n\n- system architecture diagrams\n- algorithm workflows\n- graphical abstracts\n- electronics or embedded-system schematics\n- reference-inspired redraws and layout exploration\n\nSupported conceptual-image backends:\n\n| Provider | Use case | Config |\n|---|---|---|\n| `gemini` \u002F `banana` | Google Gemini \u002F Banana-compatible generation | `NANOBANANA_*` variables |\n| `openai` | OpenAI Image API backend for conceptual figures and edits | `OPENAI_API_KEY` or `OPENAI_API_KEY_FILE` |\n\nExample:\n\n```powershell\npython \"$HOME\u002F.codex\u002Fskills\u002Fengineering-figure-agent\u002Fscripts\u002Fgenerate_image.py\" `\n  --provider openai `\n  --model gpt-image-1.5 `\n  --figure-template system-architecture `\n  --lang en `\n  --openai-quality auto `\n  --openai-size auto `\n  \"A retrieval-augmented generation system with OCR, chunking, embedding, vector search, reranking, and answer synthesis.\"\n```\n\n### Plot Mode\n\nUse `plot mode` when values, axes, and geometry must stay exact.\n\nBest for:\n\n- benchmark bar charts\n- ablation plots\n- trend curves\n- heatmaps\n- scatter plots\n- multi-panel quantitative figures\n\nRule of thumb:\n\n| If the figure needs... | Use |\n|---|---|\n| exact values, axes, error bars, or benchmark geometry | `plot mode` |\n| concept explanation, layout exploration, or architecture visuals | `image mode` |\n| both | local plot panels first, image generation only for conceptual parts |\n\n## Workflow\n\nRecommended upstream handoff:\n\n1. Use `ai-research-writing-guide` to decide what claim the figure should support.\n2. Write or collect the figure brief: audience, claim, modules, labels, data, and target style.\n3. Use `engineering-figure-agent` to render the diagram or exact plot.\n4. Verify labels, numeric truth, publication style, and export format before paper submission.\n\n## Platform Adapters\n\n| Platform | Entry | Best use |\n|---|---|---|\n| Codex | [SKILL.md](.\u002FSKILL.md) | Full agent-native workflow with local scripts |\n| Claude Code | [adapters\u002Fclaude-code\u002F](.\u002Fadapters\u002Fclaude-code\u002F) | Local repository figure work and prompt\u002Fplot generation |\n| ChatGPT \u002F Claude web | [docs\u002Fprompt-pack.md](.\u002Fdocs\u002Fprompt-pack.md) | Chat-only figure brief and prompt drafting |\n| VS Code \u002F Obsidian | [templates\u002Ffigure-brief\u002F](.\u002Ftemplates\u002Ffigure-brief\u002F) | Edit and archive figure briefs, prompts, and plot requests |\n\nCore platform-neutral contracts:\n\n- [Figure brief spec](.\u002Freferences\u002Ffigure-brief-spec.md)\n- [Plot request schema](.\u002Fschemas\u002Fplot-request.schema.json)\n- [Figure brief schema](.\u002Fschemas\u002Ffigure-brief.schema.json)\n\n## Support Matrix\n\n| Platform | Status | Notes |\n|---|---|---|\n| Windows | tested | primary tested platform; helper scripts supported first |\n| macOS | reported working | successful installs have been reported, including AI-assisted setup |\n| Linux | expected for core Python workflow | some environments may need small manual adjustments |\n\n## Repository Map\n\n| Path | Purpose |\n|---|---|\n| [SKILL.md](.\u002FSKILL.md) | Internal Codex skill instructions |\n| [providers.md](.\u002Fproviders.md) | Provider-neutral API configuration notes |\n| [references\u002F](.\u002Freferences\u002F) | Figure templates, plotting rules, prompt templates, API notes |\n| [scripts\u002F](.\u002Fscripts\u002F) | Image generation, plotting, setup checks, wizard scripts |\n| [examples\u002Ffigure-briefs\u002F](.\u002Fexamples\u002Ffigure-briefs\u002F) | Reusable figure brief examples |\n| [docs\u002Fexamples\u002F](.\u002Fdocs\u002Fexamples\u002F) | Showcase images, prompts, and notes |\n| [templates\u002Ffigure-brief\u002F](.\u002Ftemplates\u002Ffigure-brief\u002F) | Platform-neutral figure brief templates |\n| [adapters\u002F](.\u002Fadapters\u002F) | Claude Code and future platform adapters |\n\n## What It Is Not\n\n- Not a full paper-upload web platform.\n- Not a replacement for checking scientific truth, labels, and numeric values.\n- Not a single prompt that treats plots, diagrams, and schematics as the same task.\n- Not a place to commit real API keys or private provider relay details.\n\n## Notes\n\n- Keep real API keys outside the repository.\n- Prefer local plotting for exact quantitative figures.\n- Keep provider-specific private relay details out of public docs unless clearly marked as optional examples.\n- For exact publication plots, never rely on image generation for the values, axes, or benchmark geometry.\n\n## License\n\nMIT License. See [LICENSE](.\u002FLICENSE).\n","Engineering Figure Agent 是一个用于生成学术出版风格工程和计算机科学图表的Codex技能，支持中文学术论文、双语标注及可编辑图表的工作流。其核心功能包括图像模式和绘图模式，前者适用于系统架构图等概念性图表的生成，后者则适合需要精确数值的曲线图或热力图等。项目还提供了多种后端支持，如Gemini、OpenAI以及本地绘图。特别适合于撰写工程技术类论文的研究者使用，能够显著提高图表制作的效率与质量。",2,"2026-06-11 02:46:52","CREATED_QUERY"]