awesome-scientific-figures

# Awesome Scientific Figures — Skills, MCPs & Plugins for AI Coding Agents > A curated list of Skills, MCP Server, plugins, and Python libraries for generating scientific figures with AI coding agents like Claude Code and OpenAI Codex. Generate publication-quality figures, architecture diagrams, and illustrations with AI-powered tools. Say goodbye to basic matplotlib and let AI coding agents help you create stunning visualizations. ## Table of Contents - [Skills](#skills) - [Codex Exclusive Skills](#codex-exclusive-skills) - [Claude Code Exclusive Skills](#claude-code-exclusive-skills) - [Cross-Platform Skills](#cross-platform-skills) - [MCP Server](#mcp-server) - [Draw.io Series](#drawio-series) - [Excalidraw Series](#excalidraw-series) - [TikZ / LaTeX Series](#tikz--latex-series) - [General Diagrams](#general-diagrams) - [Codex Official Plugins](#codex-official-plugins) - [Python Libraries for Programmatic Figure Generation](#python-libraries-for-programmatic-figure-generation) - [Recommended Combinations](#recommended-combinations) ## Skills ### Codex Exclusive Skills | Skill | Stars | Description | |---|---|---| | [**engineering-figure-agent**](https://github.com/heyu-233/engineering-figure-agent) | ⭐245 | Dual-mode: **Image Mode** (system architecture diagrams, algorithm flowcharts, illustrations, and graphical abstracts) and **Plot Mode** (precise numerical benchmark bar charts, ablation experiment plots, trend curves, heatmaps, and scatter plots) | | [**abstract-fig**](https://github.com/keros68/abstract-fig) | New | Codex-exclusive — generates **editable draw.io graphical abstracts and manuscript concept maps**: Image2 generates elements → transparent separation PNG → embedded draw.io independent objects → text boxes/arrows/labels fully editable, output `.drawio` | | [**paper-framework-figure-studio-pro**](https://www.sourcepulse.org/projects/29689594) | New | Paper framework figure studio — specializes in generating pipeline architecture diagrams and framework illustrations for the Method section of papers | | [**archify**](https://github.com/tt-a1i/archify) | v2.5 | 5 chart types (Architecture/Workflow/Sequence/Data Flow/Lifecycle); dark/light theme SVG auto-switching; PNG/JPEG/WebP/SVG export with up to 4x resolution; single HTML zero dependency | | [**codex-skill-academic-slides**](https://github.com/FA-T-T/codex-skill-academic-slides) | New | Codex academic slides/posters/proof diagrams/architecture diagrams generation, based on GPT Image 2 | ### Claude Code Exclusive Skills | Skill | Stars | Description | |---|---|---| | [**scipilot-figure-skill**](https://github.com/Haojae/scipilot-figure-skill) | ⭐ Recommended | **"Think before you draw" philosophy**: analyze data → determine argument goals → recommend chart types; 15 interception rules to prevent bad practices; v2.1 visual self-inspection; Okabe-Ito colorblind safety color scheme | | [**academic-figure-skill**](https://github.com/TingxiYu/academic-figure-skill) | ⭐ Recommended | **29 chart types** (volcano plots, heatmaps, UMAP, violin plots, radar charts, Sankey diagrams, forest plots, etc.); targets Nature/Cell/Science standards; 4-round QA protocol (anti-pattern scanning → code compliance → visual logic → rendering verification) | | [**paper-figure**](https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep/blob/main/skills/skills-codex-claude-review/paper-figure/SKILL.md) | ⭐ Recommended | 8-step workflow: read paper proposal → environment setup → automatic figure selection → generation → batch running → LaTeX embedding → quality review → visual inspection checklist; 300 DPI, serif font, vector PDF, no title (title uses LaTeX caption) | | [**paper-illustration**](https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep/blob/main/skills/paper-illustration/SKILL.md) | ⭐ Recommended | **Architecture/method illustrations** — Claude planning → Gemini layout optimization → style verification → rendering → Claude strict review (1-10 points, iterated to ≥9); CVPR/ICLR/NeurIPS style; green input → blue encoder → purple fusion → orange output color scheme | | [**bioinformatics-plotting-standards**](https://github.com/Rooneyxu/bioinformatics-plotting-standards) | New | Bioinformatics publication-quality single plots/multi-panel combinations (volcano plots, heatmaps, UMAP/t-SNE, violin plots, dot plots, Manhattan plots, spatial transcriptome plots); built-in QC scripts | ### Cross-Platform Skills | Skill | Stars | Description | |---|---|---| | [**drawio-diagram-builder-skill**](https://github.com/Will-hxw/drawio-diagram-builder-skill) | ⭐ Recommended | Research-style editable draw.io diagrams — **screenshot-driven self-correction loop** (generate → preview → screenshot → self-inspect defects → repair → repeat); includes MIT-licensed Tabler SVG icon library; supports reference image recreation | | [**paper-illustrate**](https://skillsmp.com/zh/creators/shiien/self-evolved-research-framework/skills-paper-illustrate) | New | Generates **strictly academic-standard TikZ/SVG diagrams**: architecture diagrams, pipeline diagrams, concept maps, flowcharts, comparison charts; constrained color palette, consistent spacing, colorblind safety | | [**archify**](https://github.com/tt-a1i/archify) | v2.5 | Claude/Codex CLI/opencode universal; Mermaid dialect input; CJK font degradation; golden-file CI testing | | [**AgentFigureGallery**](https://github.com/Dsadd4/AgentFigureGallery) | 16k+ samples | 10 scientific chart types with 16,341 visual references; browser-based liking/filtering; exports reference packages for drawing code generation | | [**NaturePanelForge**](https://github.com/littlepeachs/NaturePanelForge) | New | Converts scientific paper images and Nature series open-access papers into panel-level executable drawing code reconstruction tasks | | [**diagram-to-image**](https://github.com/zju-zhanglu/diagram-to-image) | New | Converts ASCII/Unicode text diagrams in Markdown to draw.io PNG images | ## MCP Server ### Draw.io Series | MCP Server | Installation Command | Highlights | |---|---|---| | [**drawio-mcp (Python version)**](https://github.com/yohasacura/drawio-mcp) 🏆 | `uvx drawio-mcp` | **310+ shapes, 44 edge styles, 21 color themes**; C4/BPMN/K8s/network topology/ER/UML; Sugiyama DAG auto-layout; **one-click polishing**; supports multi-page/multi-layer | | [**drawio-mcp (JS version)**](https://github.com/Sujimoshi/drawio-mcp) | `npx drawio-mcp` | `.drawio.svg` format VSCode seamless compatibility; hierarchical/circle/organic/radial-tree/partition/stack auto-layout | | [**drawio-mcp-server (lgazo)**](https://github.com/lgazo/drawio-mcp-server) 🏆 | `npx -y drawio-mcp-server --editor` | **Built-in draw.io editor** (localhost:3000); auto-discovery of AWS/GCP/Azure/Cisco icon libraries; Mermaid import; multi-tab/multi-page; TLS/HTTPS | | [**@drawio/mcp (official)**](https://www.npmjs.com/package/@drawio/mcp) | `npx -y @drawio/mcp` | Draw.io official MCP; supports native XML, CSV, Mermaid syntax input | | [**AWS sample-drawio-mcp**](https://github.com/aws-samples/sample-drawio-mcp) | `npx -y https://...` | AWS official architecture diagrams (CloudFront/API Gateway/Lambda/Aurora/S3/Bedrock, etc.); 100% local execution | ### Excalidraw Series | MCP Server | Installation Command | Highlights | |---|---|---| | [**codex-excalidraw**](https://www.npmjs.com/package/codex-excalidraw) 🏆 | `npx codex-excalidraw` | Codex/Claude **directly draw on editable Excalidraw canvas**; Mermaid → Excalidraw conversion; bidirectional editing (AI drawing → human editing → AI reading canvas → improved); local workspace `http://127.0.0.1:3000/` | | [**drawmode**](https://github.com/teamchong/drawmode) | `npx drawmode --stdio` | LLM only needs to write ~10 lines of TypeScript (instead of 500 lines of JSON); Graphviz auto-layout; outputs `.excalidraw`, PNG, SVG | | [**@kamiazya/whiteboard**](https://github.com/kamiazya/whiteboard) | `npx -y @kamiazya/whiteboard-mcp@latest` | **Real-time collaborative whiteboard** — WebSocket synchronization; Claude Code/Codex/Gemini CLI universal; bundled with 3 Skills | | [**@airmix/mcp-excalidraw-server**](https://www.npmjs.com/package/@airmix/mcp-excalidraw-server) | npm install | **26 MCP tools** — element-level CRUD control; `describe_scene` + `get_canvas_screenshot` let AI "see" the canvas; snapshot/rollback; shareable URL; Docker support | ### TikZ / LaTeX Series | MCP Server | Installation Command | Highlights | |---|---|---| | [**tikz-mcp**](https://github.com/FlorianHuo/tikz-mcp) 🏆 | `pip install tikz-mcp` | Local `pdflatex` compilation of TikZ → **1000 DPI PNG** (transparent background); built-in style guide + **AI self-inspection process** (AI automatically checks and repairs output before presenting); v0.2.0 (2026.3) | | [**upmath-mcp**](https://github.com/danielsimonjr/upmath-mcp) | `npm install` | Renders TikZ through i.upmath.me API **without local TeX installation**; supports tikz-3dplot, pgfplots, circuitikz, bussproofs, mhchem | | [**texra-scientific-skills**](https://github.com/texra-ai/texra-scientific-skills) | GitHub clone | 9 scientific Skill packages, including `tikz-figure-builder` (iterative compilation + workflow viewing), `math-ocr`, `manuscript-review`, etc. | | [**Overleaf Claude MCP**](https://github.com/junfei-z/overleaf-claude-mcp) | GitHub clone | Claude Code ↔ Overleaf direct connection — read/write/push LaTeX files; Smart Diff mode saves tokens; v1.1.0 (2026.4) | | [**TeXFlow MCP**](https://glama.ai/mcp/servers/aaronsb/texflow-mcp) | `uvx texflow-mcp` | Structured document model — LaTeX is output product, not direct editing object; Markdown import | ### General Diagrams | MCP Server | Installation Command | Highlights | |---|---|---| | [**diagrams-mcp**](https://github.com/ByteOverDev/diagrams-mcp) | `uvx diagrams-mcp-server` | Based on mingrammer/diagrams; 15+ cloud providers, 500+ node types; Mermaid + PlantUML dual engines; outputs PNG/PDF/JPG/DOT | | [**graphviz-mcp**](https://www.npmjs.com/package/graphviz-mcp) | `npx -y graphviz-mcp` | Arbitrary description → Graphviz rendering; interactive viewer pan/zoom; **no separate Graphviz installation required** | | [**CC_SysML**](https://github.com/atk-pulsar/CC_SysML) | GitHub clone | Codebase → SysML v2.0 model → Gemini generates professional PNG; 15 MCP tools; supports BDD/IBD/PKG/ACT/SEQ diagrams | | [**@usetessera/mcp**](https://www.npmjs.com/package/@usetessera/mcp) | npm install | C4 model architecture documentation (Context/Container/Component/Module); 34 tools; git freshness verification | ## Codex Official Plugins The following plugins are from the [openai/plugins](https://github.com/openai/plugins) repository and are the most mature extensions in the Codex ecosystem: | Plugin | Category | Diagram/Visualization Capabilities | |---|---|---| | [**figma**](https://github.com/openai/plugins) | Design | 🏆 **Most complete reference implementation** — `use_figma` reads design → generates code, Code to Canvas reverse synchronization, Code Connect, design system rules; includes skills + commands + hooks + UI HTML | | [**canva**](https://github.com/openai/plugins) | Design | Searches for templates, creates designs, edits existing assets; social media graphics/posters/cover images; supports one-click export to PowerPoint → Canva refinement | | [**remotion**](https://github.com/openai/plugins) | Video | **Code is video** — React controls subtitles/transitions/chart animations/progress bars; localhost:3031 timeline editor | | [**google-slides**](https://github.com/openai/plugins) | Presentation | Presentation automation — programmatically creates/edits Google Slides | | [**build-web-apps**](https://github.com/openai/plugins) | Development | Complete workflow with UI components, charts, and dashboards for web visualization | | [**vercel**](https://github.com/openai/plugins) | DevOps | Largest plugin (47 skills + 3 agents + 5 commands); includes visualization dashboards for frontend deployment | | [**render**](https://github.com/openai/plugins) | DevOps | 20+ skills; includes architecture and infrastructure diagram-related deployment processes | ## Python Libraries (Programmatic Beautiful Plotting) Upgrading matplotlib's default output from the Python side: | Library | Installation | Highlights | |---|---|---| | [**dartwork-mpl**](https://pypi.org/project/dartwork-mpl/) 🏆 | `pip install dartwork-mpl` | **AI Native** - Comes with MCP Server + Prompt corpus; Physical width API (`figsize("13cm", "col1")`); Deterministic layout (arithmetic non-heuristic); OKLCH color system; `validate_figure()` auto-inspection | | [**figrecipe**](https://pypi.org/project/figrecipe/) 🏆 | `pip install figrecipe` | Millimeter-level precise layout; **Built-in diagram DSL** (boxes/cylinders/stadiums/documents/code blocks/folders) for direct architecture drawing; SCIEX/SCIEX_DARK/MATPLOTLIB presets; GUI editor | | [**Swizz**](https://github.com/) (ICML 2025) | `pip install swizz` | One-line code plotting; DataFrame→LaTeX table; Flutter-style block layout (PlotBlock/Row/Col/LegendBlock) for combining multiple panels | | [**AutoFigure**](https://github.com/ResearAI/AutoFigure) (ICLR 2026) | GitHub | **Text→Graph / Paper PDF→Graph**; Dual-agent Review-Refine loop; Outputs SVG or draw.io mxGraph XML | | [**Tesserax**](https://blog.apiad.net/p/introducing-tesserax) | `pip install tesserax` | Zero-dependency pure Python; Native SVG; **Anchor system** (relative positioning, no manual coordinates); Parameterized and reusable | | [**Netgraph**](https://github.com/paulbrodersen/netgraph) | `pip install netgraph` | Publication-level network/graph visualization; Node overlap resolution, edge crossing minimization, edge untangling; Interactive editing | | [**CACHAI**](https://pypi.org/project/cachai/) | `pip install cachai` | Specialized for chord diagrams/correlation plots - Element-wise styling, logarithmic scaling of line widths, Bezier smoothing | --- ## Recommended Combinations ### 🏆 All-Around Player (Full Coverage of Scientific Scenarios) ``` ┌─────────────────────────────────────────────┐ │ Architecture/System Diagrams/pipeline diagrams │ │ → drawio-mcp (Python version) + abstract-fig │ │ → Alternative: paper-illustration Skill │ ├─────────────────────────────────────────────┤ │ Data plots (bar/line/scatter/heat/volcano/UMAP...) │ │ → dartwork-mpl + Swizz │ │ → Alternative: scipilot-figure-skill │ ├─────────────────────────────────────────────┤ │ TikZ / LaTeX paper plots │ │ → tikz-mcp + texra-scientific-skills │ │ → Alternative: upmath-mcp (no TeX installation) │ ├─────────────────────────────────────────────┤ │ Quick whiteboard/hand-drawn style │ │ → codex-excalidraw + drawmode │ │ → Alternative: @airmix/mcp-excalidraw-server │ └─────────────────────────────────────────────┘ ``` ### 🥇 Claude Code User Minimum Viable Setup ```bash # Skills git clone https://github.com/Haojae/scipilot-figure-skill.git ~/.claude/skills/scipilot-figure git clone https://github.com/wanshuiyin/Auto-claude-code-research-in-sleep.git # Copy paper-figure and paper-illustration to ~/.claude/skills/ # MCP claude mcp add drawio -- uvx drawio-mcp claude mcp add tikz -- pip install tikz-mcp # Python pip install dartwork-mpl swizz ``` ### 🥇 Codex User Minimum Viable Setup ```bash # Skills git clone https://github.com/heyu-233/engineering-figure-agent.git git clone https://github.com/keros68/abstract-fig.git # MCP & Plugins codex mcp add drawio -- npx -y @drawio/mcp codex plugin marketplace add kamiazya/whiteboard # Python pip install dartwork-mpl figrecipe ``` --- ## 📊 Capability Matrix | Demand | Preferred Tool | Installation/Acquisition | |---|---|---| | Paper data plots (matplotlib alternative) | dartwork-mpl + Swizz | `pip install dartwork-mpl swizz` | | Architecture/System Diagrams | drawio-mcp (Python version) | `uvx drawio-mcp` | | Pipeline/Framework diagrams | paper-illustration Skill | GitHub clone | | Hand-drawn style flowcharts | codex-excalidraw / drawmode | `npx codex-excalidraw` | | TikZ paper plots | tikz-mcp | `pip install tikz-mcp` | | Cloud architecture diagrams (AWS/GCP/Azure) | AWS sample-drawio-mcp | `npx` | | Graphical abstracts | abstract-fig (Codex) | GitHub clone | | Network/graph visualization | Netgraph | `pip install netgraph` | | Editable draw.io diagrams | drawio-mcp-server (lgazo) | `npx -y drawio-mcp-server --editor` | | Scientific slides | codex-skill-academic-slides | GitHub clone | | Chord/correlation plots | CACHAI | `pip install cachai` | --- ## 🔗 Related Resources - [openai/plugins](https://github.com/openai/plugins) — Codex official plugin collection (170+) - [ComposioHQ/awesome-claude-skills](https://github.com/ComposioHQ/awesome-claude-skills) — Claude Skills curated list (64k+ Stars) - [ComposioHQ/awesome-codex-skills](https://github.com/ComposioHQ/awesome-codex-skills) — Codex Skills curated list (14k+ Stars) - [anthropics/skills](https://github.com/anthropics/skills) — Anthropic official Skills (127k+ Stars) - [SkillsMP](https://skillsmp.com) — Agent Skills marketplace and discovery platform --- ## ⚖️ License [CC0 1.0 Universal](https://creativecommons.org/publicdomain/zero/1.0/) — This list is in the public domain, feel free to use, modify, and share. --- > **Maintenance tip**: The AI tool ecosystem for scientific plotting experienced explosive growth in 2025-2026. If you discover new excellent tools, feel free to open an issue or PR!