[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-2632":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":16,"forks30d":16,"starsTrendScore":16,"compositeScore":19,"rankGlobal":10,"rankLanguage":10,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":23,"hasPages":21,"topics":24,"createdAt":10,"pushedAt":10,"updatedAt":45,"readmeContent":46,"aiSummary":47,"trendingCount":16,"starSnapshotCount":16,"syncStatus":15,"lastSyncTime":48,"discoverSource":49},2632,"AgentFigureGallery","Dsadd4\u002FAgentFigureGallery","Dsadd4","Drop-in scientific plotting skill for Claude Code, Codex, Cursor, and other coding agents.","https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002Fdsadd4\u002FAgentFigureGallery",null,"Python",128,1,60,2,0,4,24,45.3,"MIT License",false,"main",true,[25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44],"agent-skill","agent-tools","ai-agent","bioinformatics","cell-style","claude-code","codex","cursor","data-visualization","ggplot2","human-in-the-loop","llm-agents","matplotlib","nature-style","plotting-agent","reference-gallery","science-style","scientific-figures","scientific-visualization","skill","2026-06-12 04:00:15","# AgentFigureGallery\n\n[![English](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flang-English-007ec6.svg)](README.md)\n[![简体中文](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flang-%E7%AE%80%E4%BD%93%E4%B8%AD%E6%96%87-c0392b.svg)](README.zh-CN.md)\n\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-MIT-blue.svg)](LICENSE)\n[![Python 3.10+](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.10%2B-3776ab.svg)](pyproject.toml)\n[![Full KB](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ffull--public-16k%2B%20references-0f766e.svg)](docs\u002FREMOTE_FULL_VALIDATION.md)\n[![Hugging Face Dataset](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FHugging%20Face-dataset-ffcc00.svg)](https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002Fdsadd4\u002FAgentFigureGallery)\n\nAgentFigureGallery is a scientific plotting reference gallery for Claude Code, Codex, Cursor, and other coding agents.\nIt lets an agent search real figure references, lets you mark examples as liked, rejected, or selected in a browser gallery, and exports those choices as a reference bundle for plotting code.\n\n**Quick install for Codex:**\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002FDsadd4\u002FAgentFigureGallery\u002Fmain\u002Fscripts\u002Finstall.sh | bash\n```\n\nThe script clones or updates the repository at `$HOME\u002FAgentFigureGallery`, creates a Python virtual environment, installs the package, and installs the Codex skill wrapper. After this bootstrap install, run the CLI as `~\u002FAgentFigureGallery\u002F.venv\u002Fbin\u002Fagentfiguregallery`, or activate the environment first:\n\n```bash\nsource ~\u002FAgentFigureGallery\u002F.venv\u002Fbin\u002Factivate\n```\n\nFor an editable manual install, see [Manual Install](#manual-install).\n\n![AgentFigureGallery dynamic demo](docs\u002Fassets\u002Fagentfiguregallery-demo.gif)\n\n```text\nagent query -> browser gallery -> human like\u002Freject\u002Fselect -> reference bundle -> plotting code\n```\n\nBefore generating plotting code, the agent queries visual references, the user selects preferred examples in the browser, and AgentFigureGallery exports the selected references for the final plotting task.\n\n![AgentFigureGallery candidate counts by plot type](docs\u002Fassets\u002Fagentfiguregallery-scale-overview.png)\n\n## Manual Install\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FDsadd4\u002FAgentFigureGallery.git\ncd AgentFigureGallery\npython -m venv .venv\nsource .venv\u002Fbin\u002Factivate\npip install -e .\nagentfiguregallery doctor\nagentfiguregallery install-skill --target codex\n```\n\nThe default install is enough for smoke tests and the small built-in reference pack. To use the full 16k+ public reference pool, run the setup command below.\n\n## Full Public Reference Pool\n\nInstall the complete `full-public` pack from Hugging Face:\n\n```bash\nagentfiguregallery setup --pack full-public --manifest-url https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002Fdsadd4\u002FAgentFigureGallery\u002Fresolve\u002Fmain\u002Fresource_manifest.json\n```\n\nIf Hugging Face is blocked, use the GitHub API manifest fallback:\n\n```bash\nagentfiguregallery setup --pack full-public --manifest manifests\u002Fresource_manifest.github-api.json\n```\n\nYou can also download the full pack during bootstrap:\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002FDsadd4\u002FAgentFigureGallery\u002Fmain\u002Fscripts\u002Finstall.sh | env AFG_INSTALL_FULL_PUBLIC=1 bash\n```\n\n## Browser Gallery Workflow\n\nCreate a reference session and open the browser gallery:\n\n```bash\nagentfiguregallery gallery --plot-type embedding_plot --limit 50 --serve\n# Then open http:\u002F\u002F127.0.0.1:8765\u002F\n```\n\nThe command prints a `session` id before starting the local server. In the browser, mark references as liked, rejected, or selected. Those saved preferences are reused by later sessions.\n\nAfter selecting references, export the bundle for the coding agent:\n\n```bash\nagentfiguregallery bundle --session \u003Csession_id>\n```\n\nThe bundle is written to:\n\n```text\noutputs\u002Freference_sessions\u002F\u003Csession_id>\u002Fexport_bundle\u002Freference_bundle.json\n```\n\nTo reopen the frontend later without creating a new reference session:\n\n```bash\nagentfiguregallery serve --host 127.0.0.1 --port 8765\n```\n\n## Verify Codex Skill\n\nAfter installing the Codex skill, Codex can discover AgentFigureGallery as a local skill.\n\n![Codex discovered Agent Figure Gallery](examples\u002Fplot_type_examples\u002Fscreenshots\u002Fcodex-skill-discovered.png)\n\nThen ask your coding agent to run a plot-type smoke test:\n\n```text\nUse AgentFigureGallery to test your installed plotting skill. Generate one Nature-style example for each supported plot type, then export PNG\u002FPDF\u002FSVG and a combined preview.\n```\n\nThe result should look like this: one Nature-style example for every supported plot type.\n\n![AgentFigureGallery plot-type smoke examples](examples\u002Fplot_type_examples\u002Ffigures\u002Fagentfiguregallery_plot_type_examples_preview.png)\n\nSee `examples\u002Fplot_type_examples\u002F` for the runnable script, source data, and PNG\u002FPDF\u002FSVG outputs.\n\n## For Coding Agents\n\nAfter `pip install -e .` finishes, tell your Codex, Claude Code, Cursor, or other coding agent:\n\n```text\nRead skills\u002Fagent-figure-gallery\u002FSKILL.md, then use AgentFigureGallery before writing publication figure code.\n```\n\nYou can install personal skill wrappers for multiple agents:\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002FDsadd4\u002FAgentFigureGallery\u002Fmain\u002Fscripts\u002Finstall.sh | env AFG_AGENT_TARGETS=\"codex claude-code cursor\" bash\n```\n\nCursor project rules need a project path, so pass it explicitly:\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002FDsadd4\u002FAgentFigureGallery\u002Fmain\u002Fscripts\u002Finstall.sh | env AFG_AGENT_TARGETS=\"cursor\" AFG_CURSOR_PROJECT=\u002Fpath\u002Fto\u002Fyour-cursor-project bash\n```\n\nOr install wrappers manually:\n\n```bash\nagentfiguregallery install-skill --target codex\nagentfiguregallery install-skill --target claude-code\nagentfiguregallery install-skill --target cursor\nagentfiguregallery install-cursor-rule --project \u002Fpath\u002Fto\u002Fyour-cursor-project\n```\n\nCodex installs to `~\u002F.codex\u002Fskills`, Claude Code installs to `~\u002F.claude\u002Fskills`, Cursor-compatible skill installs write to `~\u002F.cursor\u002Fskills`, and Cursor project rules write to `.cursor\u002Frules\u002Fagent-figure-gallery.mdc`. See `docs\u002FAGENT_QUICKSTART.md` and `examples\u002Fagent_prompt.md`.\n\nEnd-to-end examples:\n\n- `examples\u002Fend_to_end_embedding.md`\n- `examples\u002Fgenerated_embedding_plot\u002FREADME.md`\n- `examples\u002Fbefore_after_benchmark\u002FREADME.md`\n\n## Dynamic Gallery\n\nUse the browser gallery to browse candidates by plot type, reject unsuitable references, save plot-type preferences, and export selected examples for the agent. With the `full-public` pack installed, the gallery can draw from 16,341 public visual candidates across common scientific plot types.\n\n```bash\nagentfiguregallery query --task \"Nature-style embedding map for cell atlas\"\nagentfiguregallery gallery --plot-type embedding_plot --limit 100 --serve\n```\n\n## Extend Your Gallery\n\nAgentFigureGallery can grow after install. You can ask an agent to follow the expansion guide, or add a small local reference pack yourself, then inspect the new candidates in the browser gallery.\n\nTell your coding agent:\n\n```text\nRead ExtendAgent\u002FREADME.md, then expand AgentFigureGallery for \u003Cplot type or style>. Discover high-quality public scientific plotting sources, render every useful reference as a visible preview, preserve stable candidate IDs and source license metadata, rebuild the candidate index, and report candidate counts plus private-path scan results.\n```\n\nFor manual expansion, the important rules are:\n\n1. Add only references with visible preview PNGs.\n2. Preserve stable `candidate_id`, `plot_type`, preview path, source metadata, and license attribution when available.\n3. Keep large preview packs, raw upstream repositories, private paths, and tokens out of Git.\n\nSee `ExtendAgent\u002FREADME.md` for the full expansion contract and quality gates.\n\n## Community Packs\n\nCommunity packs are the public contribution path for reusable plotting references. The base `full-public` pack remains the canonical 16k+ pool maintained by Dsadd4; community contributions land first in `community_pool\u002F`, then accepted material is periodically released as installable asset packs.\n\nContribution routes:\n\n- Open a Community Pack issue to propose public sources, plot types, or a pack idea.\n- Open a PR under `community_pool\u002Fpacks\u002F\u003Cpack_name>\u002F` using the documented schema.\n- Keep large assets out of Git; accepted packs are distributed through resource manifests.\n\nAfter a community release manifest is published, users can selectively install a community pack:\n\n```bash\nagentfiguregallery setup --pack community-latest --manifest-url \u003Ccommunity_resource_manifest_url>\nagentfiguregallery gallery --plot-type embedding_plot --limit 50 --serve\n```\n\nSee `docs\u002FCOMMUNITY_PACKS.md` and `community_pool\u002FREADME.md` for contribution rules, schemas, review gates, and install patterns.\n\n## What Is Inside\n\n- 16,341 full-public visual candidates across 10 scientific plot types.\n- Browser-gallery feedback for personal or lab-specific figure preferences.\n- A small curated minimal pack committed for instant smoke tests.\n- Codex-equipped plot-type smoke examples with PNG\u002FPDF\u002FSVG outputs.\n- Backend CLI, browser gallery, Codex skill wrapper, and agent expansion guide.\n- Stable candidate IDs, saved preferences, and export bundles for agent handoff.\n- Community pack contribution path for user-submitted plotting references and periodic asset releases.\n\n## Roadmap\n\n- [Curated Cell and Science style reference packs](https:\u002F\u002Fgithub.com\u002FDsadd4\u002FAgentFigureGallery\u002Fissues\u002F3)\n- [Faster full-public mirror for China and restricted networks](https:\u002F\u002Fgithub.com\u002FDsadd4\u002FAgentFigureGallery\u002Fissues\u002F4)\n\nCompleted:\n\n- [One-command Codex skill install](https:\u002F\u002Fgithub.com\u002FDsadd4\u002FAgentFigureGallery\u002Fissues\u002F1)\n- [Generated embedding plot from a reference bundle](examples\u002Fgenerated_embedding_plot\u002FREADME.md)\n- [Before\u002Fafter benchmark: prompt-only vs reference-guided plotting](examples\u002Fbefore_after_benchmark\u002FREADME.md)\n\n## Docs\n\nUser docs:\n\n- `docs\u002FAGENT_QUICKSTART.md`: minimal instructions for coding agents.\n- `docs\u002FCOMMUNITY_PACKS.md`: community contribution rules and release model.\n- `community_pool\u002F`: staging area and schema examples for community packs.\n- `ExtendAgent\u002F`: instructions for agents that expand the gallery.\n- `docs\u002FREMOTE_FULL_VALIDATION.md`: first remote full-public validation and current mirror-speed caveat.\n\nMaintainer docs:\n\n- `docs\u002FDISCOVERY_PLAYBOOK.md`: launch and star-growth checklist.\n- `docs\u002Freleases\u002Fv0.1.0.md`: first public release notes.\n- `docs\u002FHF_SYNC.md`: Hugging Face dataset card and asset sync commands.\n- `docs\u002FPYPI_RELEASE.md`: Python package release path.\n- `docs\u002FHF_DATASET_CARD.md`: Hugging Face dataset card draft.\n- `docs\u002FLAUNCH.md`: public launch copy and channels.\n- `docs\u002FFULL_KB_DISTRIBUTION.md`: public asset-pack strategy.\n","AgentFigureGallery 是一个为 Claude Code、Codex、Cursor 等编码代理设计的科学绘图参考库。它允许用户通过浏览器画廊浏览并标记喜欢、拒绝或选择的图表示例，然后将这些选择导出为绘图代码的参考包。项目支持多种数据可视化风格，包括类似 ggplot2 和 Nature 风格的绘图，并且能够与多个 AI 编程助手集成。适用于需要生成高质量科学图表的研究人员和开发者，在进行数据分析和报告撰写时尤为有用。","2026-06-11 02:50:34","CREATED_QUERY"]