What problem does it solve? Agents often cannot natively inspect local files such as videos, PDFs, Office documents, 3D models, or medical volumes, forcing ad-hoc shell scripting or external APIs. This Skill provides MCP tools that render and read these files directly so the agent can inspect their contents. ## Core Features & Use Cases - File visualization: Render PDFs, Office documents, CSV/XLSX data, code, notebooks, HTML pages, 3D models (OBJ/STL/GLB/STEP), GIS files, and NIfTI medical volumes into text or images the agent can inspect. - Media reading and metadata: Use media_info to inspect video/audio duration, fps, codecs, rotation, and tracks before reading; use read_image and read_video for frame extraction with resolution budgets. - Image annotation and export: Crop regions, draw bounding boxes (including grounding output from the api capability), and save document pages or video frames to files with save_view. - Use Case: Given a folder of action-cam clips and a PDF storyboard, run media_info on each clip to detect VFR and rotation issues, extract representative frames with read_video, and render the PDF pages for review — all without leaving the agent session. ## Quick Start Ask the agent to inspect the metadata of a local video file and then extract a few frames from it for review.
