Find the line.
Keep the proof.
Make PDFs searchable. Retrieve the relevant passage with its source page, then check what changed.
Local Python toolkit. Try it below with fictional documents.
Measured on this query. Characters, not tokens. Source metadata is additional.
Know what needs attention.
A saved result from the real audit command.
Duplicate bytes, missing companions and stale text are
surfaced before an agent uses them.
Read-only inspection. Nothing is automatically removed or refreshed.
The amount changed. So did the date.
Two revisions of a fictional invoice, compared by the Python CLI. Native text and form changes appear with page references; visual-only changes are flagged separately.
Pages are compared by position. This is not OCR or an exhaustive visual equivalence check.
Your files stay useful
after the chat ends.
The PDF remains the source. A Markdown companion keeps its text searchable in Obsidian or any editor. CLI retrieval checks the source hash and returns just the passage you need.
This site searches pre-extracted fictional text. PDF extraction, OCR, compression and filesystem actions run in the local CLI. No document upload.
# Inspect first. Nothing changes. scripts/pdf audit ./documents scripts/pdf batch ./documents # Create missing Markdown companions. scripts/pdf batch ./documents --apply
Python 3.11+ · Optional local OCR on macOS · Installation guide ↗