Plotting Agent

Renders matplotlib plots and conceptual diagrams from your experimental logs and research ideas, with optional vision-model critique loops to catch illegible text or misleading axes. Part of the PaperOrchestra academic paper generation pipeline, runs in parallel with literature review and produces 300 DPI PNGs plus context-aware captions. The heavy lifting happens through either a PaperBanana backbone (if you have it configured) or plain matplotlib fallbacks with academic styling baked in. Costs roughly 20 to 30 LLM calls per paper. Use this when you need publication-ready figures that follow conference template requirements, not when you're just exploring data. The hard rules around DPI, aspect ratios, and color palettes feel rigid until you've had a figure rejected by a LaTeX template.

GitHub stars: 572
last updated: Jun 2, 2026

View on GitHub

Install

$ npx -y skills add ar9av/paperorchestra --skill plotting-agent --agent claude-code

Installs into .claude/skills of the current project.

Files

Plotting Agent (Step 2)

Faithful implementation of the Plotting Agent from PaperOrchestra (Song et al., 2026, arXiv:2604.05018, §4 Step 2 and App. F.1 p.45).

Cost: ~20–30 LLM calls. The paper uses PaperBanana (Zhu et al., 2026) as the default backbone with a closed-loop VLM-critique refinement. This skill expresses that loop in host-agent terms: you (the host agent) generate matplotlib code with your own LLM, render via your Bash/Python tool, optionally critique the rendered PNG with your vision model, redraw, and finally caption.

Inputs

workspace/outline.json — specifically the plotting_plan array
workspace/inputs/idea.md and workspace/inputs/experimental_log.md — the source data
workspace/inputs/figures/ — optional pre-existing figures (PlotOn mode)

Outputs

workspace/figures/<figure_id>.png — one PNG per plotting_plan entry (300 DPI, sized to the requested aspect ratio)
workspace/figures/captions.json — {figure_id: caption_text} map