Security audit
Security checks for vulnerabilities and agentic risk
This thesis-review skill is coherent and purpose-aligned, with privacy cautions around thesis PDFs and broad trigger phrases but no evidence of hidden, destructive, or deceptive behavior.
Install only if you intend the agent to review thesis documents. Use it with theses you are authorized to process, be careful with unpublished research and personal details, and review any generated .docx files before sharing them formally.
The trigger phrases are generic academic terms such as '论文评审' and '评审意见', which can cause the skill to activate in ordinary academic conversations unrelated to formal thesis review. Over-broad invocation increases the chance that sensitive thesis content is processed unintentionally and that the skill generates structured outputs or follows its workflow in contexts the user did not intend.
The skill instructs extraction of detailed data from PDF theses and later generation of Word review documents, but it does not warn users that theses may contain unpublished research, personal information, or confidential academic material. Without an explicit notice and consent boundary, users may unknowingly expose sensitive data to analysis steps, external search comparison, or local artifact creation.
The related-files section discloses internal-looking paths and artifact names, including a memory record and archived review documents. While not directly exploitable by itself, this kind of enumeration reveals the skill's internal structure and suggests the existence of historical records that may contain sensitive academic evaluations, which can aid follow-on probing or social engineering.
| 文件 | 说明 |
|------|------|
| `skills/thesis-review-pro/SKILL.md` | 本技能文档 |
| `memory/2026-05-21.md` | 13篇论文评审记录 |
| `评审意见_*.docx` | 历史评审意见存档 |
No suspicious patterns detected.