Security audit
Security checks for vulnerabilities and agentic risk
This is a disclosed Chinese PDF-processing skill with expected document features, though users should be careful with sensitive files and irreversible-looking edits.
Install only if you want a Chinese-language PDF assistant. Avoid giving it confidential PDFs, passwords, signatures, certificates, or regulated documents unless you are comfortable with the agent processing that content, and keep backups before decryption, watermark removal, batch renaming, or signing workflows.
The skill advertises very broad trigger terms like “PDF”, “转换”, and “识别”, which are common in ordinary conversation and can cause unintended activation. That increases the chance users disclose sensitive documents or invoke file-processing behavior without clear intent, especially in chats involving contracts, IDs, invoices, or other confidential material.
The OCR functionality is described without any privacy or sensitivity warning, despite being designed to extract text from scanned PDFs and images. Users may submit highly sensitive content such as IDs, contracts, invoices, or medical/financial records without being warned that the tool will process and expose extracted text, increasing privacy and data-handling risk.
The sections covering decryption, watermark removal, and batch operations present security-impacting and potentially irreversible actions without warnings, guardrails, or authorization checks. This can facilitate removal of document protections at scale, accidental destruction of provenance markings, or mass modification of sensitive files, amplifying both misuse and user harm.
The trigger phrases listed in the description are extremely generic terms such as “PDF”, “转换”, and “识别”, which are likely to appear in ordinary user conversations unrelated to intentionally invoking this skill. That increases the chance of accidental activation on sensitive document workflows, including OCR, decryption, signing, or batch processing, which can expose private files or cause unintended integrity-impacting actions.
The skill advertises privacy- and integrity-sensitive capabilities including OCR on uploaded documents, digital signatures, encryption/decryption, watermark removal, and batch processing, but it does not warn users about data sensitivity, document authenticity risks, or irreversible changes. In practice, users may submit confidential PDFs, passwords, signatures, or regulated documents without informed consent or may unintentionally alter legal or evidentiary records.
The skill name and description are centered on a Chinese-language variant ("-cn") and all examples and interface text are presented in Chinese, while no user opt-in or language selection mechanism is documented. Under the policy, forcing a specific language or locale without user choice can be a natural-language policy violation unless clearly justified as region-specific.
No suspicious patterns detected.
