What problem does it solve? Screenshots intended for public articles or chats often contain internal file paths, account IDs, private class names, or other sensitive text, and manually guessing mask rectangles produces inaccurate or untrustworthy results. This Skill replaces ad hoc visual guessing with a script-driven workflow that calculates precise text-line coordinates before rendering the redacted image. ## Core Features & Use Cases - Script-Based Coordinate Calculation: Writes a dedicated Python script per image that detects text-line bounding boxes from pixel data using Pillow, then defines redaction targets as semantic line indexes plus x-ranges. - Deterministic Rendering: Emits JSON audit output and an ffmpeg drawbox filter string, renders the masked image with ffmpeg, and strips metadata with ImageMagick. - Editorial Redaction Policy: Treats masking as a content decision, keeping public context visible while hiding internal paths, identifiers, tokens, and secrets, with visible gray mask styling. - Iterative Adjustment: Supports "消しすぎ/もっと隠して" feedback by editing the script's REDACTIONS list and re-rendering from the original image. - Use Case: You have a terminal screenshot for a technical blog post showing internal file paths and account names. The Skill generates a coordinate script, masks only the sensitive lines, and produces a publication-safe image embedded in the thread. ## Quick Start Mask the sensitive internal information in this screenshot for public release by writing a dedicated coordinate-calculation script and rendering the redacted image with ffmpeg.
