🔍 Read the full analysis: Why Making My Vibe-Coded PDF Editor Reliable Was Worth It on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Thorsten Meyer reports that six days of testing and revisions turned Docestra, an AI-built PDF editor, from a working first version into a more thoroughly tested personal tool. Tests found issues with saved markups, encrypted files and redaction; he says version 1.5.1 had 745 unit tests. Docestra is private, and its time or cost savings have not been measured.
Thorsten Meyer says six days of testing and revisions turned Docestra, his AI-built PDF editor, from a working first version into a more thoroughly tested personal tool. In an account published November 13, he describes bugs found across 930 real PDFs, challenges with encrypted files and multiple review rounds on a redaction feature; the report matters as a case study in the work required after an AI-generated app first appears to function.
Meyer says he gave an AI coding agent a one-sentence brief on November 6, 2026, to build an Acrobat-like editor for his own use. The first version was committed that evening. It could open PDFs, add annotations and signatures, fill forms, rearrange pages and save files. The project, named Docestra, is a local TypeScript web app running in a browser on Meyer’s Mac; he says files stay on the machine.
A robustness test of 930 PDFs found that saving some files could produce a broken cross-reference table, causing markups to disappear. Meyer says he changed the save method to avoid compressed object streams. In a separate test, 52 of 69 encrypted PDFs did not permit annotations under restrictions set by their authors. The PDF library could not re-encrypt edited files, so Docestra was changed to ask for confirmation before unlocking them and to save an unprotected copy without overwriting the original. Meyer reports that all 69 test files then unlocked.
Redaction required a larger redesign. An early white-out tool covered text visually but left the underlying content in the PDF. Meyer says an independent AI review identified 14 issues in the first review round and 13 in the second for an initial real-redaction design. He replaced it with a method that converts affected pages to images, removes marked content from the pixels and adds a searchable text layer that omits text touching the redacted areas. Tests on 917 PDFs found no leaks, according to Meyer. That result is limited to his test set and is not an independent certification.
Docestra · Nov 6–13, 2026
From one prompt to a tool I can rely on
What the first version gave me, what real files broke, and what it took to fix it, from my project log.
What the tool rests on (schematic)
Built bottom up. The top layer is a plan, not a finished feature.
What happened, by date
- Nov 6One-sentence brief; first working version committed the same evening
- Nov 6Test over 930 real PDFs finds a save bug that dropped markups
- Nov 752 of 69 encrypted PDFs forbid annotations: unlock only into a new copy
- Nov 8Real redaction: review finds 14, then 13 problems; simpler design, 0 leaks in 917 PDFs
- Nov 12Version 1.5.1: 745 unit tests, 32 browser test suites
- Nov 13Version 2.0 starts: one shared PDF engine for our AI agents (in development)
Source: Docestra project records (development log, plans for versions 1 and 2.0, commit history), Nov 6–13, 2026. The layer stack is a schematic, not a measurement.
Graphic: Powered by Thorsten Meyer AI
Testing Exposed Gaps in Daily Use
The account illustrates a distinction between a software demo and a tool people can trust with routine documents. A feature may appear to work on a handful of sample files while failing on files produced by common software or carrying restrictions. Here, the consequences ranged from lost annotations to a redaction tool that could leave sensitive text recoverable beneath a visual cover.
Meyer reports that the work also changed how the editor handles encrypted documents and how it tests for regressions. Automated unit tests increased from 17 in the first version to 745 in version 1.5.1. The figures describe the project’s recorded test count, not proof that every failure mode has been covered. For readers experimenting with AI-built software, the account offers a practical caution: generating a usable first version does not by itself establish reliability, security or suitability for sensitive work.
The author says he kept improving Docestra because it already fit his workflow and because test cases could preserve lessons from earlier failures. But he has not measured time or cost savings. The project is personal and private, so the report does not establish whether the approach would produce similar results for a team, a commercial product or other users.
Products Worth Considering
Dell UltraSharp U3225QE 31.5-Inch 4K UHD 120Hz IPS Black Thunderbolt Hub Monitor
Apple MacBook Neo 13-inch Laptop with A18 Pro Chip, 8GB Memory, 512GB SSD, Liquid Retina Display, and Touch ID – Blush
Canon SELPHY CP1500 Wireless Compact Photo Printer Bundle with RP-108 Ink and Paper Set, White
As an affiliate, we earn on qualifying purchases.
From One-Day Build to Version 1.5.1
Meyer defines vibe coding as building software mainly by describing a desired result in ordinary language and having an AI model write the code. He says Anthropic’s Claude Code, running Claude Opus, wrote Docestra’s code, while OpenAI’s Codex reviewed it from the start. The initial feature list included viewing and searching PDFs, highlighting and drawing, adding notes and shapes, signing, form filling, page rotation and reordering, combining and extracting pages, printing, undo and saving.
The version history covered November 6 through November 12, moving from the first release to version 1.5.1. Meyer says the project also required decisions about third-party software licences: he excluded fonts distributed under GPL-2.0 because that licence did not meet his project rules, and later checked the licences of libraries considered for text recognition and editing. He describes testing and rework as taking longer than building for several packages, but provides no time or cost figures.
“Vibe coding gets you a working tool quickly. Turning it into a tool you can rely on is the real work.”
— Thorsten Meyer, in his November 13, 2026 account
Limits of the Reported Tests
The test results are reported by the project’s author; the source does not describe an independent audit of the full editor or its security. The finding of zero redaction leaks in 917 PDFs applies to that test set and cannot establish that all documents or use cases are safe. It is also unclear how the editor performs across operating systems, PDF producers or users beyond Meyer.
Docestra is not available to download, and Meyer says he has not measured time or cost savings. The report does not provide enough information to compare its reliability or workload with commercial PDF editors. Its version 2.0 is still in development, so its planned features and release timing remain unsettled.
Version 2.0 Remains in Development
Meyer says development of Docestra version 2.0 began November 13, 2026, the day his account was published, and that it is not finished. The source does not give a release date or say whether the software will become publicly available. For now, the next step described is continued development; any broader conclusions about its reliability, savings or suitability for other users will require more evidence.
Key Questions
What is Docestra?
Docestra is a private PDF editor Meyer built for his own use with AI coding agents. He says it runs locally in a browser on his Mac and that files do not leave the machine.
What problems did testing uncover?
Tests found that some saved PDFs could lose markups, that many encrypted test files restricted annotation, and that an early white-out feature did not remove the text underneath. Meyer says he revised saving, encrypted-file handling and redaction in response.
Does the redaction test prove the editor is safe for confidential files?
No. Meyer reports no leaks in a test of 917 PDFs, but that result covers only the files tested and is not described as an independent security certification.
Can the public download Docestra?
No. Meyer says the tool is private and not available to download. He also says version 2.0 is in development and is unfinished.
Did Docestra save time or money?
Meyer says he has not measured time or cost savings. His account describes development and testing work, but does not quantify either kind of benefit.
Source: ThorstenMeyerAI.com
