Your PDF.
Less metadata.

01 / KNOW YOUR FILE

What hidden metadata lives inside a PDF?

A PDF can look ready to share while its properties still name the person, company, or software that created it. A résumé exported from an office template might carry a previous author. A proposal may expose an internal project title. A timestamp can reveal when a document was prepared even after its visible date changes. Start by inspecting the actual file you plan to send, not an earlier version or the source document.

Document information often sits in the trailer's /Info dictionary. Familiar keys include /Title, /Author, /Subject, /Keywords, /Creator, /Producer, /CreationDate, and /ModDate. Applications can also add custom entries such as /Company, /Manager, or workstation identifiers. A fixed list of familiar keys is therefore an incomplete removal strategy. ZeroTrace enumerates the dictionary, including custom and empty entries, then removes the dictionary reference.

XMP is a separate metadata representation, usually RDF/XML in a /Metadata stream. It may carry another author value, identifiers, dates, or application-specific properties. Catalog metadata describes the document; page and other object metadata can describe smaller parts. XML does not mean the file contains readable plain text: stream filters can compress the packet. An inspector must resolve and decode supported streams before presenting a useful preview. See the PDF Association's specification archive for the underlying document structure.

Application-private data needs separate attention. /PieceInfo can hold private processing information at document or page level, including a last-modified date and vendor data. The PDF Association's custom metadata application note explains this mechanism. ZeroTrace also checks /SpiderInfo, associated with web capture information, and removes the trailer file identifier. These checks are a defined scope, not an inventory of every way information can be hidden in a PDF.

02 / UNDERSTAND THE DIFFERENCE

Why editing Properties may leave information behind

Clearing the author box in a viewer is useful housekeeping, but it does not tell you whether custom properties or another XMP copy remain. A blank dialog describes what that viewer displays. It does not establish what every PDF object contains. A name can also appear in a comment, form field, attachment, image, or the visible page itself. Those are different removal tasks.

Another distinction is how an edit is saved. PDF supports incremental updates: a newer revision can be appended while earlier bytes remain in the file. For example, ExifTool documents that its PDF metadata edits are reversible incremental updates. Seeing the new value does not establish that the previous value disappeared. That behavior is valuable for fast editing, but it matters when the purpose is removal.

ZeroTrace writes a fresh PDF after removing the supported structures and discarding unreachable objects. It then loads those new bytes and repeats the metadata inspection. The output becomes downloadable only after this check succeeds. This is a useful engineering check against accidental leftovers in the supported scope; it is not a mathematical proof of anonymity or a second implementation's forensic review.

For a sensitive release, examine the pages, search the visible text, inspect comments and attachments, and open the output in your normal viewer. Preserve your original separately until you are satisfied with the result. If a workflow requires accessibility tags, PDF/A conformance, editing features, or signed provenance, validate those requirements with the tools used by the recipient. Removing metadata is not a substitute for that acceptance check.

03 / THE LOCAL PIPELINE

How ZeroTrace processes PDFs in your browser

The browser reads the files you choose into memory. A local JavaScript worker parses each PDF, gathers the before report, removes the supported metadata structures, saves a new byte stream, and reopens it for the after report. The browser creates your download from those bytes. The processing code does not upload the selected PDF to an application server. This implementation uses JavaScript; it does not claim to use WebAssembly.

1. SelectYour PDF enters browser memory.
2. InspectReview properties and XMP previews.
3. RemoveRebuild without checked metadata.
4. VerifyReload the generated bytes.
5. DownloadSave a separate cleaned copy.

Local processing removes the need to entrust a conversion service with a copy of your document. It also uses your device's memory and processing power, so a large or complex PDF may take longer than a small export. Batch work is processed in sequence to limit resource pressure. Keep the tab open until the operation finishes, and use the clear control when you are done. Downloads remain on your device until you remove them.

The libraries are served with the site instead of fetched from a third-party CDN. That reduces external dependencies during a visit. It does not mean the entire website is invisible to the network: requesting the page still reaches the website host. If advertising is enabled in a future deployment, ad providers may make their own requests and use cookies or similar technologies according to the site's privacy disclosures and consent choices. No advertising runs in this local preview.

The before report groups likely identifying properties as higher exposure and dates, software versions, and XMP packets as medium exposure. This is a prioritization heuristic, not a probability that someone will identify you. A project name that seems harmless in one context may be sensitive in another. Read the values and consider the audience instead of relying on a badge alone.

The downloadable audit records the removal operation and the result of the reload check. Treat it as a local technical record. It is not a digitally signed attestation, legal certification, or proof of content redaction. Reports may contain information about the original document, including its filename and detected properties. Review an audit before sharing it with someone who should not receive that information.

04 / PICK THE RIGHT WORKFLOW

Command-line alternatives and ZeroTrace

A browser tool is convenient when you want to inspect a few PDFs without installing software. Command-line tools are useful for repeatable processing, local scripts, and independently checking an output. Their names are not interchangeable guarantees: compare the actual options, output behavior, and limitations for the version you use.

MAT2's documentation also makes a broader point: an empty metadata listing is not evidence that every possible metadata channel is absent. Choose the level of transformation your workflow permits. A visually rebuilt document and a structure-preserving edit make different tradeoffs. Always inspect the final file that will actually leave your device.