Security audit

她这话什么意思 · 直男聊天急救手册

Security checks for vulnerabilities and agentic risk

Overview

This is a Chinese relationship-reply coaching skill with explicit safety boundaries and no executable, account, network, or persistence behavior.

Install this if you want Chinese-language coaching for drafting respectful dating or relationship replies. Be aware it is opinionated, gender-framed, and may activate on related Chinese phrases; avoid pasting unnecessary personal identifiers, and review any suggested message yourself before sending.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

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 (5)

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The skill content is entirely Chinese-oriented and prescriptive in its response format, but it does not clearly declare a language/locale constraint or provide fallback behavior for users in other languages. This can lead to misfires where users receive guidance in an unexpected language or the system applies culturally and linguistically specific relationship advice to inputs it may not accurately understand, increasing the risk of misunderstanding in sensitive social situations.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 92% confidence
Finding: The skill advertises a long trigger list made up of broad, everyday conversational phrases, which increases the chance of accidental invocation outside the intended context. That can cause the agent to activate on unrelated personal chats and provide unsolicited interpersonal coaching on sensitive relationship content, creating privacy and UX risks even if the skill’s stated goals are non-malicious.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 91% confidence
Finding: This markdown file presents all guidance exclusively in Chinese and does not offer an alternative language or note that the skill is intended only for Chinese-speaking users. Under the language/locale policy, forcing a specific language without user opt-in can be a natural-language policy violation.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 80% confidence
Finding: The document states example responses using '她' and later repeatedly frames the target as a woman, such as '让她愿意跟你多聊'. Although the introduction says the guidance is not gender-based, the operational examples still default to a specific gender target without offering the user a choice of framing or explicitly preserving neutrality, which can conflict with language/locale-style policy requirements about forced defaults.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 77% confidence
Finding: The self-check uses '她' and distinguishes '男性朋友' as a comparison class, reinforcing a woman-as-target default. Because SQP-3 covers natural-language policy issues, this is a relevant finding where the file's wording nudges a specific language/social framing without explicit user opt-in.
Content: No source excerpt is available for this finding.