/readable — Document to Text Extraction

You have 30 PDFs you need to grep-search for specific figures before you can write your literature review.

When you need this

• You want to batch-extract an entire folder of papers into searchable .txt files for citation verification with /audit
• Some of your PDFs are scanned images with no text layer and you need OCR
• You want persistent .txt files on disk that you can grep across many papers at once
• You want to search across multiple papers for a specific number or phrase and need all source text in one searchable location

What it does — and what it won't

/readable extracts text from PDFs, Word docs, and HTML files and saves .txt files alongside the originals. For image-based scanned PDFs, it applies visual OCR — slower, but produces a readable result. It skips files that already have a .txt (run it safely on a folder without re-processing).

Note: Claude Code's built-in Read tool can render individual PDFs natively — you don't need /readable just to read one paper. Use /readable when you need batch processing, image-based OCR, or persistent .txt files for downstream grep-based citation work.

Performance: Directory mode processes files sequentially. A folder of 30 text-based PDFs takes a few minutes; image-based scanned PDFs take significantly longer per file. Run it before you need the results, not right before a deadline.

Worked example

Dario is writing his labor economics dissertation chapter. He's downloaded 31 papers on minimum wage elasticity — many of them older scanned journals. He needs to search across all of them for specific coefficient values.

/readable "papers/minimum_wage_elasticity/"

/readable processes all 31 PDFs. For the 8 image-based scanned journals, it runs OCR and flags them separately for quality review. The 23 natively digital papers extract cleanly with page headers intact. All .txt files are now on disk, ready for /audit to grep-verify every number he cites.

Try it

/readable papers/one_paper.pdf
/readable papers/
/readable papers/long_paper.pdf 1-40