screenshot-article-extractorlisted

# Screenshot Article Extractor ## Overview Turn an ordered series of article screenshots into one faithful Markdown document: read every image in order, verify readability, extract the full text, preserve the original structure (headings, bold, blockquotes, lists, tables), and save as a single `.md` file in the target directory. ## Workflow ### 1. Inventory the images, by paste order - The user pastes screenshots in reading order. File names are NOT a reliable pattern: they may look like `Clipboard_Screenshot.png`, `Clipboard_Screenshot-1.png`, or be completely random. Never sort by file name and never assume a naming rule. - The paste order is the ONLY source of truth. Number the images 1..N in the exact order they appear in the user's message. Never reorder, dedupe, or skip pages. - Use the file name only as a label to identify a specific image in reports. ### 2. Read every image (read-only pass) - Use your multimodal vision to view each image, in order. In WorkBuddy this is the built-in Read tool; in other agent environments, map this to that host's own image-reading capability. This skill never depends on a specific tool name or on any OCR library. - Extract the text content of every page into working notes, page by page. - Do NOT write any output file yet — assembly happens only after the readability check passes for ALL pages. ### 3. Readability gate — check every page (mandatory, blocks everything) For EACH image, check: 1. **Clarity**: text is