IureOCR

Leer en español

IureOCR

Turn scans and photos of documents into searchable PDFs, on your own computer.

Download for Linux · Windows · macOS

IureOCR recognizes the text in scanned PDFs and photos of documents with Tesseract and produces a PDF with searchable text, ready to upload to Iurefficient. It is one of the Iurefficient desktop apps and shares the account, keychain and session with them.

Why IureOCR

• Your documents stay on your computer. OCR runs locally with Tesseract; neither the document nor the text is sent anywhere. An Iurefficient account is only needed to save the result there.
• The PDF still looks like the original. The result keeps the scanned image and adds an invisible text layer on top, so you can search and copy, plus a .txt with the plain text.
• Whole case files in one go. A queue with per-page progress, and PDFs that already have text are skipped instead of being processed again.
• Page tools included. View, remove, keep, rotate and reorder pages without opening another program.
• Works with your AI assistant. IureOCR --mcp lets Claude, GitHub Copilot and other MCP clients run OCR and read PDFs on this computer, with no OpenRouter key.

Screenshots

The queue: one file being recognized (page 6 of 14), one queued, one done and saved to Iurefficient, one that already had text.	The recognized text of a finished document, ready to select and copy.
Page tools: mark pages to rotate, keep or remove them, or drag them into a new order.	Settings → AI assistants: connect Microsoft 365 Copilot, GitHub Copilot in VS Code or Claude Desktop.

Features

• Local OCR with Tesseract in Spanish and English (more languages if your Tesseract has them). Nothing leaves the computer: neither the document nor the text.
• Accepts scanned PDFs and images (JPG, PNG, TIFF, BMP, WEBP). The result is a PDF with the original image and an invisible text layer on top: it can be searched and copied, and the Iurefficient server indexes it without processing it again. It also leaves a .txt with the plain text.
• Skips PDFs that already have text (same rule as the server: fewer than 100 characters in the first three pages = needs recognition). It can be forced.
• Queue with per-page progress, cancel and retry. Opening a file does not start OCR by itself: it stays as "No OCR" so you can view it, edit its pages or upload it, and Recognize text (or Recognize text in N for all of them) starts it. Settings has an option to recognize files as soon as they are opened.
• Built-in viewer: continuous scroll, go to page, zoom (buttons, +/-, Ctrl+wheel) and keyboard (PgUp, PgDn, Home, End, Esc). Only the visible pages are drawn, so files with hundreds of pages open without waiting. It highlights the page being recognized and shows the resulting PDF when OCR finishes.
• Page editing with thumbnails: remove pages, keep only the selected ones, rotate them 90°, or reorder them by dragging (also "To start", "To end" and "Reverse order"). Each edit writes a new PDF next to the original ( - pages removed, - rotated, …); the original is not modified.
• Add files by dragging them onto the window, opening them with IureOCR from the system or with iureocr://ocr?file=… links.
• Save to Iurefficient: to a project (REST API) or to any folder in the document tree (WebDAV). With the IureDav drive mounted, just choose it as the output folder.
• For editing the result it suggests OnlyOffice Desktop Editors: it detects it if installed and otherwise links to its download. It is not bundled.
• English and Spanish interface: English by default, and Spanish if the operating system is set to Spanish; it can be changed in Settings → Appearance. This is independent of the text languages Tesseract recognizes.

Planned for upcoming versions (see the CHANGELOG): merging PDFs, splitting by ranges, password protection, watermark and page numbers; real image compression; rescuing unreadable pages with a vision model; the instance's list of documents pending OCR; and Tesseract bundled on macOS too.

Download and install

Get the installer from the latest release:

The installers are not code-signed, so Windows and macOS warn the first time you open them; see Code signing policy for what that means and how to continue.

Tesseract on each system

The app looks for Tesseract in this order: the IUREOCR_TESSERACT variable, the one inside the installer, the usual system paths and the PATH. Settings shows which one it found, its version and the available languages.

How it works inside

1. pdfium rasterizes each PDF page to JPEG at the chosen resolution (300 dpi by default). Images are passed as they are.
2. Tesseract receives the list of pages and produces the PDF with the text layer itself (its pdf renderer, the same one ocrmypdf uses) and the .txt.
3. The result is saved next to the original with the suffix - OCR (configurable) or in a fixed folder.

A PDF that already has readable text is not rasterized: the app says so and offers to force it.

Use from Claude, Copilot and other agents (MCP)

IureOCR --mcp starts an MCP server over stdio, with no window. Claude Desktop, Claude Code, Copilot in VS Code or any MCP client can then use the OCR and PDF tools on this computer. OCR runs here and the client's model reads the text and does the rest, so no OpenRouter key is needed. Only the text the agent chooses to read leaves the computer.

The original is never modified and paths must be absolute. Languages, resolution and output folder come from the app's Settings. The log goes to iureocr-mcp.log, next to the app's.

IureOCR --mcp-config prints what to paste into the client's configuration, with the executable's real path:

{ "mcpServers": { "iureocr": { "command": "/path/to/IureOCR", "args": ["--mcp"] } } }

In IureOCR, Settings → AI assistants detects which assistants are on the computer and connects each one the way that works for it:

• Microsoft 365 Copilot (the one almost everybody has) cannot use programs on the computer, but it can use the Iurefficient instance's MCP server. While signed in, Create access for Copilot generates an iurmcp_… token (valid for a year) and shows the server URL and the steps to add it to an agent in Copilot Studio (X-MCP-Token header). Copilot sees what the user sees in Iurefficient; to let it read a scan, save the OCR result there. Tokens are revoked from the same row.
• GitHub Copilot in VS Code: Connect opens the vscode:mcp/install?… link; VS Code asks for confirmation and saves the server in its settings. IureOCR does not write them.
• Claude Desktop: Connect adds the entry to claude_desktop_config.json without touching anything else (a .bak-iureocr copy is kept) and warns if the app has moved. Then quit and reopen Claude Desktop.

Other clients: claude mcp add iureocr -- /path/to/IureOCR --mcp in Claude Code, or the --mcp-config entry in their configuration.

Part of the Iurefficient suite

Contributing

Issues and pull requests are welcome. Good first contributions:

• Bug reports with the log file attached (see Where it stores things for its location), and the system and version you use.
• Translations: a new interface language is a new dictionary in src/lib/i18n.svelte.ts, next to the English and Spanish ones.
• Documentation: corrections and clearer explanations in this README.

To build and run the app, see Development below.

Development

npm ci
python3 scripts/descargar-pdfium.py     # pdfium library for this platform
python3 scripts/descargar-tessdata.py   # spa, eng and osd models (tessdata_fast)
npm run tauri dev

npm run check                 # frontend types
cd src-tauri && cargo check   # backend
npm run tauri build           # installers for this platform

git commit -am "v0.2.0: …"
git tag -a v0.2.0 -m "IureOCR 0.2.0" && git push --follow-tags

Where it stores things

Credentials go to the system keychain, shared with the other Iurefficient apps; the active account (instance and email) lives in <config>/iurefficient/.

Code signing policy

• Windows installers are not code-signed for now. The first time you run one, SmartScreen may show "Windows protected your PC" with an unknown publisher: choose More info → Run anyway.
• macOS: the app is not notarized by Apple. The first time, right-click (or Control-click) IureOCR in Applications and choose Open, or go to System Settings → Privacy & Security and click Open Anyway.
• Every installer is built from this repository by GitHub Actions (.github/workflows/build.yml) and published only on the releases page. Download it from there.
• What is signed: the update packages (.exe, .msi, .app.tar.gz, AppImage, .deb and .rpm) carry a minisign signature (.sig files and latest.json), and the app checks it against the public key it ships with before installing an update.
• Maintainer (commits and reviews): Eduardo Llaguno (@ellaguno).

Privacy policy

This program will not transfer any information to other networked systems unless specifically requested by the user or the person installing or operating it.

Text recognition runs entirely on the user's computer with Tesseract; documents never leave the machine. Specifically, IureOCR connects only to:

• the Iurefficient instance that the user configures, and only once an account is set up: to check the session (at start-up and in Settings), and when the user signs in, saves a document there or manages the access tokens for Copilot. Without an account it is never contacted;
• GitHub (api.github.com and the release files on github.com): at start-up, to check whether a newer release exists (it can be turned off in Settings), and when Settings is opened, to show the latest version of the other Iurefficient apps.

It collects no telemetry and no usage statistics. Credentials are stored in the operating system keychain, never in configuration files.

License

Apache License 2.0 (see LICENSE). Tesseract (Apache-2.0), pdfium (BSD-3) and the tessdata_fast models (Apache-2.0) keep their own licenses.