A PDF cleanup job often needs several operations in sequence: keep only certain pages, reorder them, rotate a subset, add page numbers, then merge the result with another PDF. Doing that once is manageable, but repeating the same steps for new files quickly becomes tedious.
PDF Pipeline Builder represents each operation as a node and connects them into one pipeline. You can preview the PDF at intermediate nodes, save the workflow as a browser-local Recipe, or export the graph as Pipeline JSON for backup and transfer. The selected PDFs are processed in the browser and are not uploaded by the app to an external server.
Build PDF processing as an ordered sequence
PDF Pipeline Builder connects PDF Input, page operations, document processing, branching and merging, and PDF Output on a canvas. A flow such as Select pages → Reorder → Page numbers → Output executes in that order.
Node order matters. Reorder → Page numbers numbers the reordered sequence, while Page numbers → Reorder moves pages that have already been numbered. Page expressions are also evaluated against the page stream that reaches that node, not always the original document order.
Combine page operations, branching, and document finishing
Version 1.0.0 includes Select / Reorder, Delete Pages, Duplicate Pages, Insert Blank Page, Rotate, Reverse Pages, Split Pages, Merge Pages, Page Numbers, Watermark, and Text Stamp. Merge Pages can combine two to six inputs in a defined order.
Page expressions such as `all`, `1-3,5`, and `3,1,2` can be used against the stream reaching a node. The pipeline model is useful when several operations need to be layered in a deliberate sequence rather than performed as isolated edits.
Preview intermediate results at any node
Instead of discovering a mistake only after the whole pipeline runs, select an intermediate node to evaluate only the upstream path needed for that point and preview the resulting PDF. Selecting PDF Output shows the final Result rather than an intermediate preview.
Preview shows up to six pages at a time with previous and next navigation. Split nodes can switch between Selected and Rest, and individual pages can be enlarged. An unrelated branch does not need to be loaded when it is not required for the node being previewed.
Use Split and Merge to process only part of the document differently
If only certain pages need rotation, Split Pages can separate Selected pages from the Rest. Send only the selected branch through Rotate, then combine the branches again with Merge Pages.
Branching makes workflows that cannot be expressed as one straight line visible on the canvas, and multiple PDF Input nodes can also be merged into a single PDF Output.
Choose between Recipes and Pipeline JSON
Save a completed pipeline as a Recipe when you want to reuse it quickly in the same browser. After choosing new PDFs, Apply to Canvas loads the Recipe for editing, while Use this Recipe runs it as a separate graph without replacing the current canvas.
Recipes store the processing graph and settings, not source PDF bytes, generated PDF bytes, staged files, source filenames, or source page counts. Pipeline JSON preserves nodes, edges, settings, positions, viewport, and output filename, making it better for backup, transfer, or Git-based storage. PDF Inputs must be selected again in either case.
| Format | Best for | PDF files |
|---|---|---|
| Recipe | Quickly reusing the same workflow in this browser | Not stored |
| Pipeline JSON | Backup, transfer to another device, or Git-based storage | Not stored |
Use PDF Organizer for one-off organization and Pipeline Builder for repeatable workflows
Browser Kitty's PDF Organizer is well suited to one-off work such as visually reordering pages, deleting pages, and merging several PDFs in a single editing session.
PDF Pipeline Builder is better when several operations need to run in a defined order and the workflow itself should be reused later. Use PDF Organizer for a one-time document cleanup and Pipeline Builder for recurring processing applied to new PDFs.
Local processing still has document and device limits
PDF Pipeline Builder embeds `pdf-lib` and Node Editor Core in the standalone HTML and blocks runtime network connections with CSP `connect-src 'none'`. Loading the site from GitHub Pages or Browser Kitty requires the initial HTML request, but the app does not transmit the selected PDF content. Preview and output use temporary local Blob PDF URLs.
Because output documents are rebuilt from pages, document-level structures such as bookmarks, attachments, forms, signatures, and outlines are not guaranteed to survive. Editing a digitally signed PDF can invalidate the signature. Large PDFs and complex branches can also use substantial device memory. Keep important originals and verify the output before relying on it.
Step by step
- Open PDF Pipeline Builder, add a PDF Input node, and choose the PDF you want to process.
- Add page-operation and document-processing nodes from the palette and connect them in execution order.
- Configure page expressions, rotation, page numbers, Watermark, and other node settings as needed.
- When only certain pages need different processing, branch with Split Pages and reconnect the streams with Merge Pages afterward.
- Select intermediate nodes and preview the PDF state at those points to check ordering and processing results.
- Connect the final page stream to PDF Output, preview the Result, and set the output filename.
- Save the generated PDF when the result is correct.
- Save the workflow as a Recipe for repeated use, and export Pipeline JSON as well when you need portable backup or transfer.
PDF Pipeline Builder
Build PDF processing as a node pipeline, inspect intermediate results, and reuse saved Recipes with other PDFs.
Tips and limitations
- The position of nodes such as Page Numbers or Watermark changes the result, so decide the intended processing order before building a long graph.
- For long pipelines, preview intermediate nodes as you build so mistakes are easier to isolate than after the entire workflow runs.
- Keep the original when digital signatures, forms, or other document-level features matter, and verify the generated PDF afterward.
Frequently asked questions
Can I use the same pipeline with another PDF?
Yes. Save the pipeline as a Recipe, then choose new files for its PDF Input nodes. The source PDFs themselves are not stored in the Recipe.
What is the difference between a Recipe and Pipeline JSON?
Recipes are for quickly reusing a workflow in the same browser. Pipeline JSON is better for backup, transfer to another device, or Git-based storage. Neither format stores PDF bytes.
Can I preview an intermediate result?
Yes. Selecting a node evaluates the upstream path needed for that point and previews the PDF state there. PDF Output is shown as the final Result.
Can only certain pages follow a different operation?
Yes. Split Pages can separate selected pages from the rest, and Merge Pages can combine the branches again after different processing.
Can I add Japanese text as a Watermark?
In v1.0.0, custom Watermark text is limited to ASCII characters. Custom font embedding for Japanese and other fonts is not supported.
Are the selected PDFs uploaded to a server?
No. Opening the app from GitHub Pages or Browser Kitty requires the initial HTML request, but PDF loading, intermediate previews, Recipe execution, and PDF generation run in the browser, and the app does not transmit the selected PDF content.