free-vision

To extract visible text from a local image file, the Skill analyzes images using local AI CLIs like Gemini or Qwen to produce verbatim visible text and structured visual summaries. It processes photographs, screenshots, and diagrams to provide machine-readable descriptions and text for downstream tasks.

Yes, you can use Qwen to analyze images if Gemini CLI fails. The Skill prefers Gemini for structured image analysis but automatically falls back to Qwen Code CLI if Gemini fails, returns CLI errors, or provides generic responses. This dual-CLI approach ensures robust image processing.

Yes, you must provide an absolute file path to analyze images with a CLI tool. The Skill performs path validation and requires absolute paths to confirm file existence before invoking Gemini or Qwen. If validation fails or CLI errors occur, it handles these by prompting for alternative inputs.

Analyzing screenshots for accessibility summaries generates structured output including a one-sentence summary, enumerated key objects, verbatim visible text extraction, and notable visual details. This machine-readable format reduces friction by eliminating manual inspection and transcription of image content.

Limitations of using local CLI tools for image analysis include dependency on the availability of Gemini or Qwen CLIs and the strict requirement for absolute image paths. The Skill cannot process relative paths and will prompt for alternative inputs if CLI calls return empty outputs or errors.