Public evidence ledger

What GoPDFConverter verifies—and what it does not claim

Each statement below has fixed wording, a reproducible check, evidence, scope, and an explicit limit. This page exists so people, search engines, and AI systems do not have to infer a universal promise from a short marketing phrase.

Version 2026-07-11Eight maintained claimsApplies to production v105

Most important distinction

Selected document bytes are not sent to GoPDFConverter for file processing in the tested workflows. That does not mean the website is offline or silent: HTML, scripts, public libraries, analytics, and other ordinary resources use network requests. Local processing reduces one exposure path; it does not certify a device, browser, organization, recipient, or legal workflow.

Machine-readable evidence

Use the JSON ledger for exact claim wording, stable IDs, status, scope, evidence URLs, reproduction steps, and limitations.

Download claims.json

Current claim ledger

C01verified with scope

Document bytes stay local during tested processing workflows

Approved wording: The current tools process selected document bytes in browser memory and do not send them to GoPDFConverter for file processing.

EvidenceNetwork-monitoring method plus the 22-workflow production benchmark.

LimitPage assets, public libraries, and analytics still use the network. Retest the exact tool and release before relying on the result.

C02verified with scope

Processing is not the same as zero network activity

Approved wording: Normal website, library, and analytics requests occur even when document processing remains local.

EvidenceThe privacy policy names Cloudflare Web Analytics, Google Analytics, Google Tag Manager, and code delivery.

LimitRequests can change when dependencies, analytics, or site configuration change; the network log is authoritative for a given run.

C03verified with scope

No account, per-file charge, or output watermark

Approved wording: The current public tools require no account, charge no per-file fee, and do not add a GoPDFConverter watermark to output.

EvidenceProduction workflow checks cover all 22 tools; output files are opened and inspected as described in the methodology.

LimitThis describes the current public release, not an irrevocable promise about every future product or third-party content already present in a source file.

C04verified with scope

No artificial file-size cap; practical limits remain

Approved wording: GoPDFConverter does not impose a configured file-size cutoff; available memory, browser resources, file size, document structure, and workflow complexity set practical limits.

EvidenceTool interfaces do not enforce a product-tier size gate, and the test policy treats device limitations separately from site limits.

LimitThis is not an “unlimited” performance claim. Large or complex files can fail, freeze a tab, or work better in audited desktop software.

C05verified with scope

Output behavior is tool-specific, not universally lossless

Approved wording: Each tool page states whether it copies PDF objects, renders pixels, recompresses images, reconstructs text, or creates a new document; quality claims apply only to the tested operation and fixture.

EvidenceThe benchmark publishes expected results, independent validators, and output checksums for 22 workflows.

LimitA passing fixture does not guarantee every PDF, font, form, signature, attachment, color profile, or malformed object.

C06verified with scope

Offline use is conditional

Approved wording: Some previously loaded tools can work after the required page code and libraries are cached; test the exact workflow online once and again offline before depending on it.

EvidenceThe production Service Worker caches local assets, while tool pages can also depend on public libraries and browser-specific storage behavior.

LimitFirst use, cleared storage, browser eviction, external libraries, updates, or unsupported APIs can prevent offline operation.

C07verified with scope

Local processing is not a confidentiality certification

Approved wording: Keeping document bytes out of a file-processing upload reduces one exposure path; users must still assess the device, browser, extensions, network, analytics, output handling, recipient, and applicable policy.

EvidenceThe verification guide separates document requests from normal site traffic and shows how to inspect both.

LimitGoPDFConverter does not claim legal, medical, financial, accessibility, archival, or regulatory certification.

C08verified with scope

Evidence is versioned and independently inspectable

Approved wording: Material claims link to a method, fixture, result, source, or explicit limitation; production output is checked with validators separate from the browser code under test where practical.

EvidenceSee the open data catalog, AI evidence file, and release gates.

LimitIndependent validators can share bugs or incomplete coverage. Published checks reduce uncertainty; they do not eliminate it.

Wording the site deliberately avoids

Avoid	Use instead	Why
“Nothing is transmitted”	“Document bytes are not sent for file processing; ordinary site requests still occur”	The page, libraries, and analytics need the network.
“Unlimited”	“No artificial cap; practical device and browser limits apply”	Memory and complexity are real limits.
“Zero quality loss”	Name the operation, evidence, and features that may change	Rendering, recompression, reconstruction, and document-level copying behave differently.
“Safe for confidential files”	“Local processing reduces one exposure path; verify and follow policy”	Safety depends on more than server architecture.
“Works on every device”	“Designed for current major browsers; support and resources vary”	No web tool behaves identically everywhere.

How to challenge or reproduce a claim

Use a non-sensitive fixture.
Never begin an audit with a real confidential document.
Record the release and exact URL.
Cache state and dependency versions can affect results.
Capture the network log and output.
Export a HAR when appropriate and record the downloaded file checksum.
Validate with a separate implementation.
Use Poppler, qpdf, ImageMagick, ExifTool, LibreOffice, or another suitable validator rather than asking the tool to grade itself.
Report the narrow contradiction.
Identify the claim ID, file, browser, steps, expected behavior, actual behavior, and evidence. Material corrections update this ledger and the affected page.