Security audit
Security checks for vulnerabilities and agentic risk
This skill is a coherent meme-chat behavior guide with no executable code or network behavior, though users should opt in before allowing saved or passively collected meme libraries.
Install only if you want the agent to handle memes conversationally. Before enabling library features, decide whether meme saving is allowed, turn off passive collection unless you explicitly want it, and periodically review or delete <workspace>/memes/ and meme-index.md.
The explicit activation examples for image-analysis mode are all in Chinese, and the response-style examples are also exclusively Chinese. This creates a natural-language policy concern because the skill appears to privilege a specific language without opt-in or stating that it is intended only for a Chinese-language environment.
The skill explicitly permits saving user-provided memes and passively encountered images, plus maintaining usage metadata such as last-used timestamps and counts, without requiring notice, consent, or retention limits. Even though it advises avoiding private or sensitive content, this still creates a privacy risk because images and interaction metadata can reveal personal preferences, relationships, and potentially identifying context if stored by default.
No suspicious patterns detected.
