| name | ocr-tool |
| description | Extract raw text from PDF files with OCR, including full-document extraction, selected page ranges, and reuse of cached pages. Use when a user asks to OCR a PDF, extract text from a scanned PDF, or read selected PDF pages. |
OCR Tool
Use this skill only to extract source text from PDFs.
Do not add verification steps unless the user explicitly asks.
Session OCR Cache
Run the cache command directly from your current working directory.
- Default output:
.ocr-tool-cache/output.txt.
- Use
--output PATH when a specific output file is better for the task.
- Exit code
0: read all of the output file.
- Exit code
3: no valid text was extracted.
- Exit code
1 or 2: stop and report the failure.
Readiness
- Run
pdfocr --help.
- If
pdfocr is missing or fails to start, read
references/pdfocr-install.md, install the
binary or missing runtime dependency, and retry.
- After installing
pdfocr, tell the user how to configure
DEEPINFRA_API_KEY or config.json before running OCR.
Execution
- Request unrestricted network or escalated execution directly when running
pdfocr. Do not run a sandboxed network probe first.
- Do not inspect environment variables, shell profiles, or filesystem files to
discover API keys. If authentication fails, ask the user to configure the
key, then retry.
Process PDF Input
Extract text exclusively through pdfocr shell execution.
- Do not read PDFs with direct file readers or ad hoc parsers when this skill is active.
Usage
- Full document:
python3 <SKILL_PATH>/scripts/ocr_cache.py INPUT.pdf
- Page ranges:
python3 <SKILL_PATH>/scripts/ocr_cache.py INPUT.pdf --pages "8-20,22-27"
- Custom output:
python3 <SKILL_PATH>/scripts/ocr_cache.py INPUT.pdf --output ocr.txt
Clean OCR Text
Before returning extracted text, remove only clear metadata, such as:
- Headers and footers
- Page numbers
- Timestamps
- Extraneous document identifiers
Preserve all substantive content. If a fragment is too broken to recover confidently, omit it rather than guess.
