What problem does it solve?
Reading complex academic figures is time-consuming, especially when the underlying plotting code and context are unclear. This Skill interprets one or more paper figures by combining visual understanding, automatic retrieval of the source code that generated each figure, and optional user notes, producing a structured Markdown report that teaches you how to read each figure.
Core Features & Use Cases
- Multi-source figure interpretation: Converts figures to JPG for visual analysis, automatically searches nearby directories for the generating source code (.R, .py, .ipynb, .tex, etc.), and merges both into a structured report with sections for core meaning, variable definitions, key takeaways, explanation, and scientific value.
- Process-isolated parallel analysis: Runs each figure through an isolated
codex exec or claude -p subprocess with a configurable concurrency limit (default 3), keeping all intermediate files inside a dedicated hidden workspace directory.
- Read-only safety guarantee: Never modifies your figures or source code; runtime directories (HOME, TMP, XDG) are redirected into the task workspace and leaks are audited and cleaned automatically.
- Use Case: You have three result plots from a bioinformatics experiment. Point the Skill at their absolute paths, add a note about what trends you care about, and receive a single
paper-explain-figures_report.md explaining each figure's axes, variables, and scientific implications.
Quick Start
Ask the AI to use the paper-explain-figures skill to interpret your figure by providing its absolute file path, optionally with a note describing what aspects of the figure you care about.
