Datasets:
pretty_name: AgentFigureGallery
language:
- en
tags:
- scientific-figures
- data-visualization
- bioinformatics
- ai-agents
- agent-skill
- codex
- codex-skill
- claude-code
- cursor
- coding-agents
- human-in-the-loop
- nature-style
- cell-style
- science-style
license: mit
AgentFigureGallery Full Public KB
Full public visual reference pack for AgentFigureGallery, a human-preference memory system and installable agent skill for scientific plotting agents.
The pack contains 16,341 public visual candidates across 10 scientific plot types, plus metadata needed by the AgentFigureGallery CLI, browser gallery, and Codex / Claude Code / Cursor-compatible skill workflow.
Use
git clone https://github.com/Dsadd4/AgentFigureGallery.git
cd AgentFigureGallery
pip install -e .
agentfiguregallery install-skill --target codex
agentfiguregallery setup --pack full-public --manifest-url https://huggingface.co/datasets/dsadd4/AgentFigureGallery/resolve/main/resource_manifest.json
agentfiguregallery gallery --plot-type embedding_plot --limit 50 --serve
One-command Codex bootstrap:
curl -fsSL https://raw.githubusercontent.com/Dsadd4/AgentFigureGallery/main/scripts/install.sh | bash
Optional agent installs:
agentfiguregallery install-skill --target claude-code
agentfiguregallery install-skill --target cursor
agentfiguregallery install-cursor-rule --project /path/to/your-cursor-project
Candidate Counts
- bar_chart: 1,516
- benchmark_performance: 2,310
- box_violin_distribution: 1,026
- embedding_plot: 1,273
- heatmap_matrix: 1,122
- line_chart: 1,035
- microscopy_panel: 1,756
- multi_panel_figure: 4,148
- scatter_plot: 1,489
- spatial_map: 666
Intended Use
This dataset is intended for visual reference retrieval, human preference selection, and agent-guided scientific figure generation.
Recommended loop:
agent query -> gallery display -> human like/reject/select -> export bundle -> agent action
It is not intended as a benchmark for biological conclusions, image classification, or clinical decision making.
Source And Filtering
The full-public pack is filtered to remove private local paths and private repositories. Public source attribution is preserved in candidate metadata when available.