Security audit
Security checks for vulnerabilities and agentic risk
This is a prompt-only research and writing skill that transparently helps turn topic research into Chinese article materials.
Install this if you want Chinese-language research briefs and article-planning support for this topic area. Be aware that it defaults to Chinese output and may guide the agent toward English and Chinese source searches; ask explicitly for another language or narrower source scope if needed.
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
- "platforms selling emotional support"
- "care and control in media archetypes"
## Output rules
Always follow these rules:
The document explicitly defines only 'Core English queries' and 'Core Chinese queries', which imposes specific language choices in the skill content. Under the policy, language or locale constraints should either be user-selectable or clearly justified as region-specific; neither is present here.
The description says the skill converts material into Chinese-language writing material, which imposes a specific language by default. The file does not indicate that users can choose another language or opt into this locale constraint.
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.
- original interviews, official docs, conference talks, or creator statements when relevant
2. **Conceptual precision**
Never reduce the whole topic to simplistic labels like:
- “恋母”
- “缺爱”
- “就是 Daddy issues”
The skill hard-codes Chinese as the default output language unless the user explicitly asks otherwise. This can override user preference or broader system expectations, reducing transparency and potentially causing unsafe transformations if the agent silently rewrites content into another language. In this file, the behavior is editorial rather than overtly malicious, but it is still a real policy/UX control issue.
No suspicious patterns detected.