All the PDF tools you actually need, in one tab.

A free, 100% on-device PDF toolkit that bundles the ten operations people ask for most: merge, split, rotate, compress, page manager, PDF to images, images to PDF, encrypt, OCR, and watermark. It exists because every popular online PDF service uploads your file to a server, holds it for some retention window, and counts it against a daily quota — exactly the wrong shape for a tax return, a confidential contract, a client deliverable under NDA, a school transcript, or a medical record. Here the same operations run locally on the bytes you already have on your disk, and your file never leaves the browser tab.

Three open-source libraries do the work. pdf-lib handles structural edits (merge, split, rotate, delete, reorder, encrypt, watermark, image embed). pdf.js renders pages to canvases when you need images, page thumbnails, or input to OCR. Tesseract.js, compiled to WebAssembly, runs the text recognition step on the same rendered canvases — also in the browser. There is no API endpoint in the pipeline; the only network requests are the one-time fetch of each WASM blob the first time a feature needs it, after which the browser cache keeps it.

What each mode does, and when to pick it

Merge preserves every page from every source PDF in the order shown in the queue, and writes one combined PDF with the first file's metadata. Reorder the queue by drag-and-drop to interleave pages from different documents before merging — common when assembling a packet from several contracts, or pulling chapters from different drafts. Split operates the other direction: by ranges (1-3, 5, 7-9) you name the output sections yourself; by fixed size you slice the document into evenly sized chunks. Rotate is a single 90°/180°/270° pass applied to selected pages — useful after scanning a stack of mixed-orientation documents. Compress rebuilds the PDF with object streams, rewrites the cross-reference table, and optionally strips Author/Producer metadata, which usually shaves a few percent even from already-compressed files.

The Page Manager is the only operation that exposes the page grid as a visual tool. Every page renders as a thumbnail; tap to mark for deletion, hover the bottom edge to drag-reorder, then save once. PDF to Images turns pages into PNG (lossless) or JPEG (smaller) at 1×/1.5×/2×/3× scale (72/108/144/216 DPI) and packages them as a ZIP. Images to PDF is the inverse: drop a stack of PNG/JPEG/WebP/AVIF/GIF/BMP files, pick A4/Letter/A3/A5/Legal and portrait or landscape, and the tool lays each image on its own page with configurable margins.

OCR: turning scans into searchable text

OCR mode is the only one that ships a model. Tesseract.js runs as WebAssembly entirely in the browser tab; pdf.js renders each page at 2× scale (~144 DPI, the sweet spot between accuracy and memory), the canvas is passed to the Tesseract worker, and the recognized text is collected per page with a confidence score. Choose the recognition language from a list of twelve (English, simplified/traditional Chinese, Japanese, Korean, German, French, Spanish, Italian, Portuguese, Russian, Arabic). The first run downloads the matching language data file (~10–20 MB) and caches it for every later run.

Use page ranges to limit the work: "1-5, 12" runs OCR on six pages rather than the whole document, and the free tier caps a single run at five pages. Pro raises that to 600 pages per batch. The output is one block of text per page, joined with page headers, ready to copy or download as a UTF-8 .txt file. The text is plain — there is no translation, layout reconstruction, or handwriting recognition. For images that are not PDFs, use the standalone OCR tool which works directly on image files.

Encryption and password protection

Encrypt mode adds a 128-bit RC4 user password to the file. Recipients must enter it to open the document; without it, every PDF reader refuses to display content. Owner password is optional — leave it blank to use the same password for both, or set a different one if you want to grant a recipient the ability to open but not to print, copy, or change restrictions later. The password itself is never transmitted: the encryption step happens entirely in your browser, and the resulting file is a standard PDF 1.4–2.0 file any reader can verify.

Two caveats. The tool cannot open a PDF that already requires a password — there is no password-removal feature, by design. And the underlying engine implements RC4 rather than AES, because pdf-lib's AES path is not yet available in WASM. RC4 128-bit is the format the broadest set of PDF readers accept (including every reader on iOS, Android, Windows, macOS, and Linux), and is more than enough for a personal or small-team use case. If you need AES-256, open the file in Adobe Acrobat or your OS print-to-PDF flow after the operations here.

What the tool does not do

There is no AI chat, no form-field editor, no signature or annotation tool, no PDF/A conversion, no redaction engine, and no server-side anything. Files that already require a password cannot be opened. The watermarking mode is text-only, laid out diagonally across every page with adjustable size, opacity, rotation, and color — it is a stamp, not a hand-drawn signature. OCR works on a text layer produced by Tesseract; badly skewed, very low-resolution, or stylized fonts will not transcribe cleanly.

There is no daily limit on the free tier. The numbers you see in the plan (50 MB per file, 10 PDFs per merge batch, 30 pages for image export, 5 pages per OCR batch) are single-operation caps, not quotas: you can run merge once, then split, then rotate, then download, as many times as you like, in any order. Pro raises the same caps to 200 MB / 50 PDFs / 600 pages / 600 pages and unlocks higher-fidelity 3× rendering for print-quality image export.

Privacy, security, and offline use

Everything in the tool runs in the browser tab: parsing, modification, re-encoding, encryption, OCR, and image rendering. There is no upload endpoint, no telemetry that contains file content, and no account is required. The only network activity is the one-time download of pdf.js's worker script (hosted on jsDelivr, cached by the browser) and the Tesseract language data (hosted on the official jsDelivr CDN for tesseract.js). You can confirm this in the Network panel of your browser's devtools — drop a file, run a merge, watch the only requests.

After the first run the tool keeps working offline. pdf.js, the Tesseract engine, and the language data files for every language you have used all sit in the browser's HTTP cache; opening the page on a flight, in a coffee shop with no Wi-Fi, or behind a strict corporate firewall is the same experience as at your desk. Closing the tab discards every file, every thumbnail, and every intermediate result from memory — there is nothing to clean up, no temporary directory, no upload queue.

Limits, formats, and licensing

Free tier: 50 MB per file, 10 PDFs per merge, 30 pages per image export, 5 pages per OCR batch, unlimited runs. Pro: 200 MB per file, 50 PDFs per merge, 600 pages per image export, 600 pages per OCR batch, higher-fidelity rendering. A single file is always accepted — these caps only apply to the multi-file modes (merge) and to per-operation outputs (page count for image export and OCR).

Outputs are standard PDF 1.4–2.0 files with object streams, so they open in every reader, and the metadata strip is opt-in (it is on by default in compress mode, off everywhere else). PNG/JPEG output from PDF to Images is byte-identical to what pdf.js renders, then optionally re-encoded to JPEG at the quality you pick. Image inputs to Images to PDF are embedded at their native pixel dimensions, scaled to fit the page with a contain layout. Everything is yours, with no watermark, and commercial use is included on both the free and Pro plans. The underlying libraries are open source (pdf-lib MIT, pdf.js Apache 2.0, Tesseract.js Apache 2.0); their licenses are linked in the page footer.