PDF Translation Reserving Exact Same Layout
PassAudited by ClawScan on May 10, 2026.
Overview
The skill is a coherent PDF translation workflow with local helper scripts and a clearly disclosed last-resort external fallback, with no evidence of hidden exfiltration or destructive behavior.
This appears safe for ordinary PDF translation workflows. Before using it on confidential documents, avoid or explicitly approve any hosted fallback upload, and make sure local dependencies like pdftotext/poppler and pypdf are installed from trusted sources.
Findings (4)
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 skill may fail until local dependencies are installed, and users need to choose trusted package/system sources themselves.
The artifacts declare no installation requirements, while the included workflow relies on pdftotext and the script imports pypdf; this is a dependency/documentation gap rather than suspicious behavior.
Required binaries (all must exist): none ... No install spec — this is an instruction-only skill.
Document required dependencies such as pdftotext/poppler and pypdf, and install them only from trusted package managers.
The workflow runs local helper commands against the PDF the user selects.
The helper executes a local PDF text extraction command. It is purpose-aligned and uses an argument list rather than a shell, but users should still understand that installing the skill enables local command execution when invoked.
subprocess.run(["pdftotext", "-layout", "-f", str(page_number), "-l", str(page_number), str(pdf_path), "-"], check=True, capture_output=True, text=True)
Use this only with PDFs you are comfortable processing locally, and keep local PDF tooling patched.
Sensitive PDFs could leave the local environment if the fallback service is used.
The hosted fallback is disclosed and scoped as last-resort, but using it for document translation may share PDF content with an external service.
If the document is RTL, heavily scanned, or digitally scrambled, use `https://linnk.ai/doc-translator` only as the last resort.
Ask for explicit user confirmation before uploading private documents to the fallback service, and avoid the fallback for confidential PDFs unless the user accepts that data flow.
PDF text may be shared across subagents or worker contexts during parallel translation.
Parallel delegation is explicitly conditional on the user's request and is scoped by batch/page range, but it can distribute document text to other agent workers.
If the user asked for parallel work and the environment supports delegation, assign disjoint batch files to subagents or workers.
Use parallel delegation only when appropriate for the document sensitivity, and keep batches disjoint as instructed.