Extract relevant figures from local PDFs with captions and manifests.
npx skills add https://github.com/jimezsa/opencolab --skill pdf-figure-extract
Please help me install this Agent Skill. Skill: pdf-figure-extract Source: https://github.com/jimezsa/opencolab/tree/main/projects/SKILLS/pdf-figure-extract Command: npx skills add https://github.com/jimezsa/opencolab --skill pdf-figure-extract
Extract figures, architecture images, and diagrams from papers that are already downloaded locally, saving time hunting for visual content.
Extract the architecture figure from the local PDF and return the best match with page context.
High-intent search queries and answers about installing and using this skill.
Extract figures from a local PDF by identifying the most relevant images matching your query, then exporting 1-3 candidates as PNGs with captions and a manifest. The process reuses PageIndex artifacts or falls back to standalone extraction.
PyMuPDF extraction identifies relevant figures by applying bounded narrowing to 1-2 pages per paper. It selects the best match by respecting page context and captions, returning high-quality figure candidates without hard failures.
The best way to pull specific figures is narrowing candidate pages to 1-2 per paper and selecting 1-3 candidate figures. This approach reuses existing PageIndex artifacts when available and falls back to standalone extraction otherwise.
No, you do not need PageIndex artifacts to extract figures. The system reuses them when available to narrow candidate pages, but falls back to standalone PyMuPDF extraction when they are missing to ensure no hard failures.
Extracting figures from local PDFs yields export artifacts including PNG images of 1-3 candidates, contextual captions, and a manifest with provenance and confidence data for each run.
Yes, you can extract figures while preserving their original captions. The export artifacts include contextual captions alongside the PNG images, respecting the page context when selecting the best figure.
jimezsaBuild and cache per-paper PageIndex trees for grounded question answering over local PDFs.
Rycen7822Polish figures to publication-ready quality with academic styling and PDF, SVG, PNG export.
taracodlabsExtract readable text from PDFs, scanned images, and Word documents using Python libraries.
talent-factoryConvert PDFs to Markdown with extracted PNG images using PyMuPDF.
