Trust Center · Verification
Controlled launch tests for all 10 Kukureku PDF tools
Before public launch, each current tool was exercised with a controlled browser test designed to make page order, text preservation, rotation, watermarking, encryption, or output size easy to verify. These results document what was actually tested.
Verification scope
Pre-launch browser verification completed in October 2026
The tests below used intentionally simple markers and controlled fixtures so the expected output could be checked directly. They complement automated linting, unit tests, production builds, and deployment checks, but they do not guarantee that every possible PDF or Office file will behave identically.
Tool-by-tool verification results
| Tool | Controlled test | Observed result | Status |
|---|---|---|---|
| Word to PDF | A controlled DOCX containing headings, normal paragraphs, lists, a four-column table, and a long-cell table. | Converted into a three-page PDF with the expected readable content, table structure, and final marker present. | Pass |
| Merge PDF | Two controlled two-page PDFs marked A1/A2 and B1/B2. | Produced one four-page PDF in the expected A1 → A2 → B1 → B2 order. | Pass |
| Split PDF | Pages 2 and 3 extracted from the controlled four-page merged PDF. | Produced a two-page PDF containing A2 → B1 in the requested order. | Pass |
| Compress PDF | A two-page, image-heavy PDF of about 11.06 MB using the Medium compression option. | The controlled output was about 192.63 KB, roughly a 98% reduction, while retaining two readable pages. | Pass |
| PDF to Word | A controlled PDF containing a title, paragraphs, a table, post-table text, a second page, and final marker. | The DOCX preserved the expected title, paragraphs, table, text after the table, page-two content, and final marker. | Pass |
| JPG to PDF | Two controlled images with different orientation: one landscape and one portrait. | Created a two-page PDF in the selected order with the expected image orientation and no meaningful crop. | Pass |
| PDF to JPG | The controlled two-page image PDF was converted to JPG output. | The page-one landscape image and page-two portrait image were exported in the expected order. | Pass |
| Organize PDF | Starting from A1 → A2 → B1 → B2, A2 was deleted, B1 moved, and B1 rotated 90 degrees. | The saved PDF contained B1 → A1 → B2, with the intended B1 rotation retained. | Pass |
| Watermark PDF | Text watermark applied to a three-page controlled PDF at 42 pt, 25% opacity, 45-degree rotation, centered. | The watermark text appeared on all three pages, including the rotated page. | Pass |
| Unlock PDF | A genuinely AES-128 encrypted, two-page PDF was processed using its known password. | The resulting PDF was no longer encrypted and both pages remained readable with the expected markers. | Pass |
What these tests do not prove
A passing controlled fixture is not a promise that every real-world document will convert perfectly.
PDFs can contain damaged objects, unusual fonts, unsupported encryption, huge images, complex forms, scans, or layout constructs that stress a browser differently.
Compression results vary dramatically by source file. The 98% figure above belongs only to the stated image-heavy test fixture and should not be treated as a typical guarantee.
How to verify Kukureku yourself
For privacy behavior, use a harmless test document and watch the browser Network panel while running a tool. For output behavior, add obvious page labels or final markers so you can inspect the downloaded result yourself.
Read the private-processing verification guide →Ongoing verification
Kukureku's repository uses automated validation for code quality, unit tests, and production builds. Browser-level checks remain important for workflows that depend on downloads, rendered pages, document ordering, or device-side processing.
View the public source repository →