Caspenda PDF

How it works

Caspenda PDF uses a shared responsive workspace with focused browser-local tools. Each tool has a defined support boundary rather than a blanket claim of universal PDF compatibility.

Metadata Check

The read-only inspector parses supported structures in your browser with self-hosted PDF.js. It distinguishes Found, Not found in checked locations, Partially checked and Not checked. It does not claim to inspect every possible place sensitive information can exist, and it does not rate a PDF as safe.

Merge PDF

The current merge slice accepts multiple ordinary PDFs, applies combined file/page limits, refuses structures outside the supported preservation scope, and re-opens the merged result to compare page count, geometry, annotation/link summaries, and available extracted text before download is enabled. Source document-level metadata is not claimed as preserved.

Split PDF

The current split slice extracts selected pages or ascending ranges such as 2,4-6 into one new PDF. Duplicate, descending and out-of-range selections are rejected. Complex PDFs outside the supported preservation scope are refused. The output is re-opened and checked for page count, geometry, annotation/link summaries and available extracted text before download is enabled.

Organize PDF

The current vertical slice supports page reordering for ordinary PDFs. It refuses several complex structures instead of silently modifying them. After export, the result is re-opened and checked for page count, geometry and extracted text order where text is available before download is enabled.

Protect and Unlock PDF

Protect PDF and Unlock PDF load a self-hosted qpdf WebAssembly engine only when needed. Protect adds an AES-256 opening password to an unencrypted PDF; Unlock requires the current opening password and does not attempt password cracking or recovery. Processing runs in a dedicated browser Worker, outputs are verified before download, and the Worker is terminated after use.

Image to PDF

The current image slice accepts JPG, PNG and WebP, normalizes browser-decoded orientation, and creates one A4 page per image. Fit keeps the whole image visible. Fill is optional and explicitly crops edges. File-count, byte and decoded-pixel limits are enforced before conversion. The generated PDF is re-opened to verify page count, A4 geometry and image paint presence; pixel-level visual equivalence is not yet claimed.

Tool state is preserved

Each tool keeps its current file selection while you switch tabs, so a tab change does not silently discard selected files.

What remains outside the current scope

Metadata removal, XFA editing, password recovery, permission/certificate encryption, and broad full-fidelity PDF editing remain outside the current supported scope until their coverage, preservation and resource limits are established.