Lyonite

PDF Metadata Viewer

Read the document properties, every XMP property, and the EXIF and GPS inside embedded photographs — the data a properties clean leaves behind. Nothing is uploaded.

Runs on your device. Nothing is uploaded.

Drop a PDF here

PDF · processed on this device

It carries a photograph with GPS coordinates, a comment naming its reviewer, an embedded spreadsheet and a script — the things a document-properties clean leaves behind.

Where a PDF keeps its metadata

File → Properties in any reader shows the eight fields in the first row. The rest of this table is also in the file, and clearing the first row does not touch it.

Where metadata is stored in a PDF, what each location can contain, and whether clearing the document properties removes it
Where it livesWhat it can containCleared with document properties?
Document Information dictionaryTitle, Author, Subject, Keywords, Creator, Producer, and two timestampsYes
XMP metadata packetThe same fields again, plus document and instance IDs, an edit history, and whatever namespaces the producing software inventedUsually
Inside embedded photographsEXIF, IPTC and XMP blocks: GPS coordinates, camera make and model, capture time, the photographer, the agencyNo
On annotationsThe name of whoever wrote each comment, and when they wrote itNo
Embedded file attachmentsWhole files — spreadsheets, images, source documents — with their own names, sizes and timestampsNo
Earlier saved revisionsA PDF saved incrementally appends each version rather than replacing it, so previous states of the document remain in the fileNo

Lyonite reads all six and reports them separately, so a file whose properties are empty is not described as clean. We gave seven removers the same file and measured what each one left behind.

What can the viewer inspect?

A PDF can contain descriptive metadata and structural information that is not obvious from the visible pages. Lyonite combines document properties with a structural scan so you can review both in one place.

  • Document Info: Title, Author, Subject, Keywords, Creator, Producer, and dates
  • XMP: common descriptive fields, document IDs, instance IDs, language, and PDF/A declarations
  • Document structure: PDF version, file size, page count, page size, forms, and linearization
  • Security clues: encryption status and signature-related structures
  • Resources: fonts with embedding status and detected image objects
  • Interactive content: detected JavaScript actions and tagged-PDF markers

How to interpret the results

Provenance clues: Creator, Producer, Author, and timestamps can help explain how a file was produced, but they are editable metadata and are not proof of authorship or chronology.

Privacy review: Names, software identifiers, document IDs, and custom properties may reveal information you did not intend to share.

PDF/A: Lyonite reports a PDF/A level only when it is declared in XMP. That is not a full standards-conformance validation.

Accessibility: A tagged-PDF marker is useful context, but it does not prove that the document is fully accessible or correctly tagged.

PDF Info Dictionary vs. XMP metadata

A PDF may contain a Document Information dictionary, an XMP metadata stream, both, or neither. The Info Dictionary contains familiar fields such as Title, Author, Subject, Keywords, Creator, Producer, and dates.

XMP is XML-based and can carry overlapping descriptive fields plus identifiers, language information, metadata timestamps, PDF/A declarations, and application-defined properties. Lyonite reads the active XMP stream when it can decode it and exposes the common fields it understands.

XMP is extensible, so a document can contain custom namespaces that Lyonite does not currently surface individually. The viewer should therefore be treated as a practical inspection tool, not an exhaustive forensic parser.

Creator vs. Producer — what's the difference?

Creator

Usually the application that authored the source document, such as Microsoft Word, Adobe InDesign, or another document-authoring tool.

Producer

Usually the software, library, or PDF engine that generated the final PDF bytes.

These values are useful clues, but software can set or rewrite them. Treat them as metadata, not cryptographic evidence.
Read the full Creator vs. Producer guide →

Understanding digital-signature detection

Lyonite distinguishes between a signature field and a structure that appears to contain signature bytes. It does not validate the certificate chain, revocation status, timestamp authority, or cryptographic validity of a signature.

Important: Editing or cleaning a PDF rewrites the file and invalidates existing signatures over the original bytes.

How the inspector reads a PDF

Most online metadata viewers show you six fields from one place. Here is what this one does instead, and why the difference matters.

Two sources, kept apart. A PDF can describe itself twice: in the Document Information dictionary, and in an XMP packet. The two disagree more often than you would expect — an editor updates one and not the other, so a file claims two different authors depending on which a reader trusts. Tools that merge them into one list hide exactly the discrepancy worth seeing. This one reports both, and shows which value each field came from.

The object graph, not a text search. Fonts and images are found by walking the PDF's indirect objects rather than by pattern matching the raw bytes. Regex-scraping a PDF is fast and wrong: it picks up objects from superseded revisions and reports deleted content as present. Composite Type0 fonts keep their descriptor on a descendant CIDFont rather than on the font itself, so the inspector follows that reference before deciding whether a font is embedded — which is the difference between "this document carries its fonts" and "this document will substitute on someone else's machine".

A signature field is not a signature. Many documents contain an empty signature field that was never signed. The inspector separates the two: a genuine signature has both a /ByteRange and signature /Contents, and anything less is reported as a field rather than as a signed document. It does not claim to validate trust — only to tell you which of the two you are looking at.

What it deliberately will not tell you. When a PDF has been edited through incremental revisions, stale XMP packets from earlier versions can remain in the file as orphaned bytes. Scanning for those and reporting what they say would produce a richer-looking result and a misleading one: those values are not the document's current metadata. The inspector reports the XMP the catalog actually points at, and nothing else.

Beyond document properties. The report also counts what a document-properties clean would leave behind — images carrying their own EXIF, annotations naming an author, embedded attachments. A file whose properties are all empty can still carry every one of those, so an inspector that stopped at the document layer would be telling you a file is clean when it is not. That case, and how we got it wrong ourselves, is written up in why ExifTool says it is clean.

Frequently Asked Questions

How do I check the metadata of a PDF?

Select the PDF above and Lyonite will inspect its document properties and supported structural data. You can also use desktop tools such as Adobe Acrobat or command-line tools such as exiftool for a second opinion when a document requires deeper analysis.

Can PDF metadata reveal who created a document?

Sometimes. Author names, creator software, producer software, timestamps, identifiers and custom fields can provide useful context. Because metadata can be edited or removed, it should not be treated as definitive proof of who created a file or when an event occurred. See how to interpret PDF authorship clues.

What does this find that other metadata viewers do not?

Three things, all of which survive a document-properties clean and none of which appear in a typical viewer: EXIF and GPS blocks inside embedded photographs, author names and timestamps on annotations, and embedded file attachments. A PDF whose properties are entirely empty can still carry all three, which is why they are reported separately. We gave seven removers the same file: six of six left the GPS in.

What happens with password-protected or encrypted PDFs?

Lyonite blocks encrypted PDFs in the metadata workflow because the PDF engine it uses does not decrypt them for safe editing. Remove the password or encryption from a file you are authorized to access, then inspect it again.

What does the PDF version number mean?

The PDF header declares a specification version such as PDF 1.4, 1.7 or 2.0. It is a compatibility clue, not a promise that the file uses every feature from that version. Changing only the declared header does not convert document features or guarantee compatibility with older readers.

Is my file uploaded when I use this viewer?

No. The inspection runs locally in your browser, and the page is served with a Content-Security-Policy permitting network connections only back to lyonite.com, so a request carrying your document elsewhere is refused by the browser rather than caught in review.

Does a PDF/A badge mean the file is fully compliant?

No. Lyonite reports PDF/A when the active XMP metadata declares a part and conformance level. A full PDF/A validator checks many additional document requirements, so the badge is an identification clue rather than a conformance certificate.

I found sensitive data — how do I remove it?

Use Remove PDF Metadata to strip document properties, image EXIF, annotation authorship, attachments and scripts in one pass. Use Edit PDF Metadata instead when you want to change specific fields rather than clear them.

Related tools