smart_ocr
PassAudited by ClawScan on May 1, 2026.
Overview
This is a straightforward OCR instruction skill, with the main caveats being local file/code access and reliance on external OCR libraries.
This skill appears benign for OCR. Before installing or using it, be aware that it may process the files you give it through local code and third-party OCR libraries; avoid unnecessary sensitive documents, approve URL-based or bulk processing carefully, and use trusted sources for any required Python packages.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent may run OCR code against files you provide and may create temporary files while processing scanned PDFs.
The skill declares local computer, code execution, and file operation tools. These are expected for OCR over local images and PDFs, but they can read and process user files if invoked.
tools: [computer, code_execution, file_operations]
Use it only on files you intend to process, and review before allowing broad batches, URL-based inputs, or sensitive documents.
If PaddleOCR or related libraries are not already available, installing them from an untrusted source could introduce avoidable risk.
The instructions depend on an external OCR package, while the artifact set has no install spec or pinned dependency versions. This is purpose-aligned, but users may need to verify any dependency installation separately.
from paddleocr import PaddleOCR
Install PaddleOCR and related dependencies only from official or trusted package sources, and prefer pinned versions in managed environments.