SMF·ADOBE-ACROBAT-PRO
Can a prompt replace Adobe Acrobat Pro?
PDF & documents — PDF authoring, OCR and compliance output
Exhibit tracking slip
Verdict
Acrobat is two products on one bill. The first is a toolbox — merge, split, compress, OCR, convert — and open source covers it completely. The second edits the body text of a PDF somebody else authored, with the original fonts and correct reflow, and emits tagged PDF/UA and validated PDF/A. That half is decades of typesetting and standards work, and no personal build gets near it.
Exhibit A — The prompt
Received on31.07.2026Build a self-hosted PDF workbench for one person.
Deploy Stirling PDF behind a reverse proxy with HTTP basic auth, and put a small wrapper in front of it that does three things Stirling does not: it keeps an immutable copy of every uploaded original in a content-addressed store keyed by SHA-256, it writes each operation to a new file rather than overwriting, and it records an append-only log line per operation with the input hash, the operation and its parameters, the output hash and a timestamp.
Support the operations you actually use: merge, split, reorder, rotate, compress, watermark, redact, OCR with Tesseract, and Office-to-PDF via LibreOffice headless.
Add one conformance step that most tools skip: after any operation, run veraPDF against the output and store the verdict alongside it. When the file is not valid PDF/A or PDF/UA, say so in plain words on the result page. Never label an output "archival" or "accessible" on the strength of having produced it — only on the strength of the validator having passed it.
Out of scope, deliberately: editing existing body text, generating tagged PDF/UA structure, e-signatures, and any cloud account. Editing text inside a PDF someone else laid out is a font-substitution and reflow problem, not a feature you bolt on, and a build that pretends otherwise produces documents that look wrong on somebody else's screen.
Finish by running the operations end to end on a scanned document, a form and a slide deck, and print the log.Opening prefills the prompt — press enter to run it.
Exhibit B — What you lose
- B.1 editing the body text of a PDF you did not author, with fonts and reflow intact
- B.2 tagged PDF/UA output that passes an accessibility checker
- B.3 PDF/A archival output that passes a conformance validator
- B.4 Acrobat Sign and the audit trail behind it
- B.5 the Microsoft Office, Outlook and browser plug-ins
Prior art
- Stirling PDFLicense: MIT
Exhibit C — Why people still pay: rendering fidelity and compliance output
Because in a lot of jobs the deliverable is not "a PDF" but "a PDF that passes the checker" — an accessibility audit, an archival deposit, a regulator's intake form. Acrobat is the tool whose output those checkers were written against.
Questions
Can I open my existing Acrobat files in this?
Yes — they are ordinary PDFs, and nothing here changes that. What does not travel is anything stored in an Acrobat cloud account: shared review comments, saved Sign templates, and Document Cloud storage links.
Will the OCR be as good as Acrobat's?
On clean 300 dpi scans of printed text, Tesseract is close enough that you will not notice. On low-contrast phone photos, skewed pages, tables and handwriting it is noticeably worse, and you will end up re-scanning documents you would previously have thrown at Acrobat.
What does it cost to run?
A small VPS, around $5 a month, or nothing at all if you run it on a machine you already own. Stirling PDF, Tesseract, LibreOffice and veraPDF are all free and everything runs locally — there is no per-document charge.
What is the one thing that does not survive the rebuild?
Compliance output you can defend. If someone requires a tagged, accessible PDF or a PDF/A-2b archive, this build can tell you that your file is not one — but it cannot produce one.
Related tools
Receipt