Security audit

销售救场教练

Security checks for vulnerabilities and agentic risk

Overview

This is a sales-coaching prompt skill that analyzes user-provided customer conversations and does not install code, run commands, persist data, or request external access.

Install this if you want Chinese-language sales coaching on conversations you intentionally provide. Redact private customer details before pasting chats, especially phone numbers, names, company identifiers, prices, or confidential business terms.

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 (3)

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The trigger text uses broad natural-language examples like '帮我看看这个客户' and '客户嫌贵怎么办', which can overlap with ordinary conversation and cause the skill to activate outside a clearly bounded scope. Over-broad activation increases the chance of prompt routing mistakes, unexpected handling of sensitive pasted chats, and user confusion about when this skill should run.

Vague Triggers

Medium

Confidence: 93% confidence
Finding: The catch-all condition '贴任意客户聊天记录' allows activation on any pasted customer chat record without requiring the user to request sales coaching or acknowledge the content may contain sensitive data. This broad trigger makes unintended ingestion of personal or business-confidential conversations more likely and can route unrelated chats into a sales-analysis workflow.

Natural-Language Policy Violations

Low

Confidence: 90% confidence
Finding: The display names, summary, detailed instructions, output format, and trigger phrases are all specified in Chinese, and the skill instructs the model using Chinese-only examples and phrasing. There is no opt-in language choice or explicit documentation that the skill is intentionally limited to Chinese-speaking users.

Static analysis

No suspicious patterns detected.