Figures4papers Playbook
Use this skill to find the closest figures4papers example and convert it into a user-specific plotting script quickly.
Workflow
- Parse user intent into chart type and domain (for example: grouped bar + benchmark comparison).
- Run
scripts/example_locator.pyto shortlist candidate scripts. - Read
references/example_index.mdfor folder-level context and naming patterns. - Choose one base script and adapt data, labels, and file outputs.
- If style consistency is required, co-use
scientific-figure-profor final polishing.
Locator Script
Use keyword and chart filters:
python skills/figures4papers-playbook/scripts/example_locator.py --keyword bar --chart-type grouped_bar
Output includes:
- 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
After locating a base script, provide:
- Selected template path.
- What to replace (data arrays, labels, colors, limits).
- Command to run.
- Expected output files (
.pngand.pdf).