Security audit

彩虹屁·妈妈夸夸卡

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed Chinese-language content-generation skill for parenting praise cards, with local saving only after user confirmation.

Install this only if you want a Chinese-language skill that handles parenting experiences and drafts content that may be posted publicly. Review outputs for child, family, health, location, or other private details before posting, and decline the save prompt if you do not want files written locally.

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Vague Triggers

Medium

Confidence: 86% confidence
Finding: The trigger list includes short, common conversational phrases such as '夸一下' and similar praise-seeking language that can appear in ordinary chat without a clear intent to invoke the skill. This raises the risk of accidental activation, causing unsolicited generation of emotional-support content or file outputs when the user did not explicitly request the skill.

Vague Triggers

Medium

Confidence: 82% confidence
Finding: The README advertises '被动情绪识别自动提议' without defining precise activation thresholds, signals, or safeguards. Ambiguous passive detection can cause the skill to infer emotional state from routine text and intervene unexpectedly, which is especially risky in a sensitive emotional-support context involving mothers and personal experiences.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The override phrases include very common conversational expressions such as '开心一点' and '温暖一点', which can appear in ordinary chat unrelated to an active praise-card workflow. Although the skill adds a guard that overrides only apply after the praise flow is active and require confirmation, broad triggers still raise the risk of unintended mode changes, confusing outputs, and user-manipulation of downstream generation behavior.

Vague Triggers

Medium

Confidence: 85% confidence
Finding: The partial-override rules are underspecified and rely on broad natural-language phrases like '极简风', which are overloaded and can map to either output format or image style. Without stricter disambiguation and explicit constraints, the adapter may apply the wrong override or produce inconsistent combinations, enabling prompt-level steering by ambiguous user wording.

Static analysis

No suspicious patterns detected.