Security audit
Security checks for vulnerabilities and agentic risk
This is a simple local idea-tracking skill, with some usability and disclosure gaps but no evidence of hidden access, exfiltration, or unsafe execution.
Installers should understand that ideas may be saved locally and can be deleted by the skill. Use explicit commands when recording or deleting ideas, and confirm the actual storage path before relying on backups or privacy assumptions.
The trigger phrases are very broad everyday expressions such as '灵感', '发现', and '想法', which can easily match normal conversation and cause unintended activation. In this skill, unintended activation is more concerning because the documented behavior includes persistent local writes, status changes, and deletion-related operations on a user data file.
The skill explicitly stores records in a persistent local path and supports deletion behavior, but it does not clearly warn users that their content will be written to disk or explain the difference between soft delete and permanent deletion. This creates privacy and integrity risk: users may disclose sensitive ideas without realizing they are being persisted, and may also unintentionally lose data through destructive operations.
The only user-facing output is a Chinese-language message, which imposes a specific language without any opt-in or alternative. This can violate language/locale policy when the skill is not clearly documented as region-specific.
No suspicious patterns detected.
