engineering-figure-agent
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.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/heyu-233--engineering-figure-agent)HTML
<a href="/listings/heyu-233--engineering-figure-agent"><img src="/api/listings/heyu-233--engineering-figure-agent/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Install and run the setup check:
& "$HOME/.codex/skills/engineering-figure-agent/scripts/install_and_test.ps1" -RunSetupCheck
Open the wizard:
& "$HOME/.codex/skills/engineering-figure-agent/scripts/wizard.ps1"
Or generate a conceptual figure directly:
python "$HOME/.codex/skills/engineering-figure-agent/scripts/generate_image.py" ` --figure-template system-architecture ` --lang en ` "A retrieval-augmented generation system with OCR, chunking, embedding, vector search, reranking, and answer synthesis."
For installation details, use the language-specific guides:
- [README.zh-CN.md](./README.zh-CN.md): Windows 最短安装路径、验证方法、中文示例 - [README.en.md](./README.en.md): setup, verification, examples, positioning
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
SKILL.md · Manifest · 97 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
- ---
- name: engineering-figure-agent
- description: Use when the user needs engineering or research-paper figures: system architecture diagrams, algorithm workflows, hardware schematics, benchmark charts, ablation plots, figure briefs, redraws, image edits, or exact publication-style plots from numeric data. Covers Gemini, Nano Banana, OpenAI image models, and local plotting workflows.
- ---
- # Engineering Figure Agent
- Use this skill after the figure goal is already reasonably clear.
- ## Boundary
- Good fit:
- - Turn a figure brief into a conceptual diagram, workflow figure, schematic, or exact publication plot.
- - Choose between `image`, `plot`, and `mixed` mode.
- - Build prompts, render plots, and refine existing briefs.
- Not the main tool for:
- - Deciding from scratch what claim the paper should visualize.
- - Full paper-argument or reviewer-style critique.
- If the user is still deciding the figure claim, panel logic, or caption argument, use an available research-writing or paper-analysis skill upstream first.
- ## Core Decision
- - `image` mode: conceptual figures, architecture diagrams, algorithm workflows, graphical abstracts, schematics, reference-inspired redraws.
- - `plot` mode: exact bar charts, trend curves, heatmaps, scatter plots, ablation plots, benchmark summaries.
- - `mixed` mode: render quantitative panels locally first, then use image generation only for conceptual panels.
- Never use image generation for exact values, axes, or benchmark geometry.
- ## Default Workflow
- 1. Inspect the user input and decide whether a figure brief is already present.
- 2. If needed, create a brief using `references/figure-brief-spec.md`.
- 3. Choose `image`, `plot`, or `mixed` mode.
- 4. For conceptual figures, prefer the prompt-builder scripts; read template references only when you need to inspect or customize template wording.
- 5. For exact plots, create a concise plot request and render it locally.
- 6. Keep labels short, claims source-grounded, and outputs publication-oriented.
- 7. Save prompt/spec/output paths when files are produced.
- 8. Before finishing, run only the checks needed for the touched path.
- ## Reference Loading
- Read only what is needed:
- - `references/figure-brief-spec.md`: brief structure and mode rules.
- - `references/publication-figure-design.md`: publication styling defaults.
- - `references/provider-selection.md`: Gemini/Banana/OpenAI setup.
- - `references/highres-policy.md`: high-resolution and fail-closed rules.
- - `references/chinese-labels.md`: Chinese label readability rules.
- - `references/engineering-figure-templates.md` and `references/materials-science-figure-template.md`: template wording when script output needs customization.
- - `references/publication-chart-patterns.md`, `references/natural-language-plot-workflow.md`, and `references/publication-plot-api.md`: exact or plot-heavy workflows.
- - `references/editable-figure-handoff.md`: optional editable SVG handoff.
- ## Scripts
- - Build a prompt without network calls:
- ```bash
- python scripts/build_engineering_figure_prompt.py --figure-template system-architecture --lang en "technical background"
- python scripts/build_materials_figure_prompt.py --materials-figure mechanism-figure --lang en "scientific background"
- ```
- - Generate or edit a conceptual image:
- ```bash
- python scripts/generate_image.py --provider openai --figure-template system-architecture --lang en "technical background"
- python scripts/generate_image.py --provider openai --materials-figure mechanism-figure --lang en "scientific background"
- ```
- - Render an exact plot:
- ```bash
- python scripts/build_plot_spec.py request.json --out spec.json
- python scripts/plot_publication_figure.py spec.json --out-path output/figure --formats png pdf svg
- ```
- - Use the unified CLI:
- ```bash
- python scripts/efa.py prompt --figure-template system-architecture "technical background"
- python scripts/efa.py prompt --materials-figure mechanism-figure "scientific background"
- python scripts/efa.py plot request.json --out-path output/figure
- python scripts/efa.py check
- ```
- ## Quality Rules
- - Do not fabricate measurements, benchmark values, hardware specs, or unsupported causal claims.
- - Keep arrows, reading order, and module hierarchy explicit.
- - Favor white backgrounds, publication-style spacing, and readable labels.
- - Use prompt-first workflows when layout fidelity matters.
- - For high-resolution requests, follow `references/highres-policy.md`.
- - For third-party providers, require explicit opt-in before sending keys or user files.
- - For Chinese figures, preserve standard technical notation when it improves clarity.
README.md · README · 192 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
- # Engineering Figure Agent
- <div align="center">
- 
- 
- 
- 
- 
- [](https://socialistic.ai/zh/skill/engineering-figure-agent-407b80?utm_source=github&utm_medium=readme&utm_campaign=20260519-student-research-tools&utm_content=badge)
- **Agent-native figure production for engineering and CS papers.**
- 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。
- [中文说明](./README.zh-CN.md) | [English Guide](./README.en.md) | [Example Gallery](./docs/examples/README.md) | [Showcase](./docs/showcase.md)
- </div>
- ## Preview
- | System / Architecture | Cooperative Perception | Safety Taxonomy |
- |---|---|---|
- |  |  |  |
- | Dense Systems Diagram | Deployment Scenario |
- |---|---|
- |  |  |
- | Exact Local Plot |
- |---|
- |  |
- ## Why It Exists
- Most figure tools treat every paper figure as the same image prompt problem. Engineering papers do not work that way.
- | Figure need | Better path |
- |---|---|
- | System architecture, pipeline, schematic, graphical abstract | `image mode` with an engineering-aware prompt template |
- | Benchmark curves, ablation bars, heatmaps, scatter plots | `plot mode` with local plotting and exact values |
- | Mixed conceptual + quantitative figure | Render numeric panels locally, then compose or describe the conceptual panels |
- | Reference-inspired redraw | Use image mode for structure and style exploration, then manually verify labels and layout |
- Engineering 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.
- ## Quick Start
- Install and run the setup check:
- ```powershell
- & "$HOME/.codex/skills/engineering-figure-agent/scripts/install_and_test.ps1" -RunSetupCheck
- ```
- Open the wizard:
- ```powershell
- & "$HOME/.codex/skills/engineering-figure-agent/scripts/wizard.ps1"
- ```
- Or generate a conceptual figure directly:
- ```powershell
- python "$HOME/.codex/skills/engineering-figure-agent/scripts/generate_image.py" `
- --figure-template system-architecture `
- --lang en `
- "A retrieval-augmented generation system with OCR, chunking, embedding, vector search, reranking, and answer synthesis."
- ```
- For installation details, use the language-specific guides:
- - [README.zh-CN.md](./README.zh-CN.md): Windows 最短安装路径、验证方法、中文示例
- - [README.en.md](./README.en.md): setup, verification, examples, positioning
- ## Two Modes
- ### Image Mode
- Use `image mode` when visual structure matters more than exact numeric geometry.
- Best for:
- - system architecture diagrams
- - algorithm workflows
- - graphical abstracts
- - electronics or embedded-system schematics
- - reference-inspired redraws and layout exploration
- Supported conceptual-image backends:
- | Provider | Use case | Config |
- |---|---|---|
- | `gemini` / `banana` | Google Gemini / Banana-compatible generation | `NANOBANANA_*` variables |
- | `openai` | OpenAI Image API backend for conceptual figures and edits | `OPENAI_API_KEY` or `OPENAI_API_KEY_FILE` |
- Example:
- ```powershell
- python "$HOME/.codex/skills/engineering-figure-agent/scripts/generate_image.py" `
- --provider openai `
- --model gpt-image-1.5 `
- --figure-template system-architecture `
- --lang en `
- --openai-quality auto `
- --openai-size auto `
- "A retrieval-augmented generation system with OCR, chunking, embedding, vector search, reranking, and answer synthesis."
- ```
- ### Plot Mode
- Use `plot mode` when values, axes, and geometry must stay exact.
- Best for:
- - benchmark bar charts
- - ablation plots
- - trend curves
- - heatmaps
- - scatter plots
- - multi-panel quantitative figures
- Rule of thumb:
- | If the figure needs... | Use |
- |---|---|
- | exact values, axes, error bars, or benchmark geometry | `plot mode` |
- | concept explanation, layout exploration, or architecture visuals | `image mode` |
- | both | local plot panels first, image generation only for conceptual parts |
- ## Workflow
- Recommended upstream handoff:
- 1. Use `ai-research-writing-guide` to decide what claim the figure should support.
- 2. Write or collect the figure brief: audience, claim, modules, labels, data, and target style.
- 3. Use `engineering-figure-agent` to render the diagram or exact plot.
- 4. Verify labels, numeric truth, publication style, and export format before paper submission.
- ## Platform Adapters
- | Platform | Entry | Best use |
- |---|---|---|
- | Codex | [SKILL.md](./SKILL.md) | Full agent-native workflow with local scripts |
- | Claude Code | [adapters/claude-code/](./adapters/claude-code/) | Local repository figure work and prompt/plot generation |
- | ChatGPT / Claude web | [docs/prompt-pack.md](./docs/prompt-pack.md) | Chat-only figure brief and prompt drafting |
- | VS Code / Obsidian | [templates/figure-brief/](./templates/figure-brief/) | Edit and archive figure briefs, prompts, and plot requests |
- Core platform-neutral contracts:
- - [Figure brief spec](./references/figure-brief-spec.md)
- - [Plot request schema](./schemas/plot-request.schema.json)
- - [Figure brief schema](./schemas/figure-brief.schema.json)
- ## Support Matrix
- | Platform | Status | Notes |
- |---|---|---|
- | Windows | tested | primary tested platform; helper scripts supported first |
- | macOS | reported working | successful installs have been reported, including AI-assisted setup |
- | Linux | expected for core Python workflow | some environments may need small manual adjustments |
- ## Repository Map
- | Path | Purpose |
- |---|---|
- | [SKILL.md](./SKILL.md) | Internal Codex skill instructions |
- | [providers.md](./providers.md) | Provider-neutral API configuration notes |
- | [references/](./references/) | Figure templates, plotting rules, prompt templates, API notes |
- | [scripts/](./scripts/) | Image generation, plotting, setup checks, wizard scripts |
- | [examples/figure-briefs/](./examples/figure-briefs/) | Reusable figure brief examples |
- | [docs/examples/](./docs/examples/) | Showcase images, prompts, and notes |
- | [templates/figure-brief/](./templates/figure-brief/) | Platform-neutral figure brief templates |
- | [adapters/](./adapters/) | Claude Code and future platform adapters |
- ## What It Is Not
- - Not a full paper-upload web platform.
- - Not a replacement for checking scientific truth, labels, and numeric values.
- - Not a single prompt that treats plots, diagrams, and schematics as the same task.
- - Not a place to commit real API keys or private provider relay details.
- ## Notes
- - Keep real API keys outside the repository.
- - Prefer local plotting for exact quantitative figures.
- - Keep provider-specific private relay details out of public docs unless clearly marked as optional examples.
- - For exact publication plots, never rely on image generation for the values, axes, or benchmark geometry.
- ## License
- MIT License. See [LICENSE](./LICENSE).
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.