Security audit
Security checks for vulnerabilities and agentic risk
The skill is a simple paper-editing prompt with no code or persistence, but it explicitly markets anti-AI-detection rewriting for academic work.
Review carefully before installing. The skill has no executable code and does not request elevated system access, but users should not use its anti-AI-detection guidance to hide AI assistance, misrepresent authorship, or bypass journal, university, funder, or employer disclosure rules. Treat unpublished manuscripts and document paths as sensitive input.
The title and description present the skill entirely as a Chinese-language assistant for paper editing, and nowhere in the document is the user offered a language or locale option. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the constraint is explicitly justified.
The skill explicitly instructs users to 'eliminate AI writing traces,' which is direct semantic support for evading AI-origin detection. Even though framed as academic polishing, this enables concealment of machine-generated authorship and can facilitate academic integrity violations, deceptive submissions, or policy circumvention.
Describing the skill as suitable for users seeking 'anti-AI-trace' writing makes evasion a marketed use case rather than an incidental side effect. In the academic-paper context, this is more dangerous because it encourages deceptive representation of authorship in submissions subject to journal, university, or funding-body disclosure rules.
The 'AI detection standards' section operationalizes concrete heuristics for making text appear less AI-generated, which amounts to practical detection-evasion guidance. This crosses from general editing into a repeatable method for disguising AI-assisted content, especially risky in academic publishing where provenance and disclosure may be compliance requirements.
