Figures4papers Playbook

Workflow

1. Parse user intent into chart type and domain (for example: grouped bar + benchmark comparison).
2. Run

   scripts/example_locator.py

   to shortlist candidate scripts.
3. Read

   references/example_index.md

   for folder-level context and naming patterns.
4. Choose one base script and adapt data, labels, and file outputs.
5. If style consistency is required, co-use

   scientific-figure-pro

   for final polishing.

Locator Script

python skills/figures4papers-playbook/scripts/example_locator.py --keyword bar --chart-type grouped_bar

• Folder and domain.
• Matching script paths.
• Recommended run commands.
• Notes about the intended figure purpose.

Selection Rules

• Prefer exact chart-type match first.
• Then prefer domain match (biomed, LLM eval, vision survey, etc.).
• If no exact match exists, choose the simplest script with nearest visual grammar.
• Keep output paths under a dedicated

  figures/

  directory in the target project.

References

• references/example_index.md

  : curated index of available figure folders and scripts.

Handoff Pattern

1. Selected template path.
2. What to replace (data arrays, labels, colors, limits).
3. Command to run.
4. Expected output files (

   .png

   and

   .pdf

   ).