What is dococr?

docOCR is a Mac command-line tool that reads document images and converts them to Markdown text. You give it one or more image files, and it produces text output that preserves the document structure: paragraphs, lists, and tables are formatted as proper Markdown rather than plain unstructured text. All processing happens locally on your Mac, no images are sent to any external service. The tool relies on Apple's document recognition API, which is only available on macOS 26 or later. It uses Apple's built-in machine learning to identify the layout and content of a document from a photo or scan. There are two ways to use it. The command-line mode accepts a list of image files. By default it prints the recognized text to the terminal. With an optional flag it writes a Markdown file next to each source image using the same filename with a .md extension. The server mode starts a local web server on your machine. You can then open a browser, upload an image through a simple web page, and see the recognized text. The same server also exposes a JSON API so other programs can send an image and receive the OCR result as structured data. A practical example shown in the README is a macOS Shortcuts workflow. You capture a screen selection with a screenshot tool, the shortcut passes the image to docOCR, reads the Markdown text from the output, and copies it to your clipboard. You can then paste it into any text editor. Building the tool requires Xcode or the Swift command-line tools. The README includes steps for building a release binary and installing it to a directory on your system path. An optional companion skill for the Codex AI assistant is also available as a separate repository.