Security audit

Taobao Shopping

Security checks for vulnerabilities and agentic risk

Overview

This Taobao shopping skill is a read-only shopping assistant that clearly keeps login, cart, checkout, orders, and payment under the user's control.

Before installing, understand that the skill may browse public Taobao pages and summarize product, seller, review, price, and promotion information. Do not provide passwords, SMS codes, payment details, address details, or account data to the agent; complete cart, checkout, order submission, and payment yourself.

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 (4)

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The usage example shows the user invoking the skill with “帮我买蓝牙耳机,” which is a generic shopping request and common everyday phrasing. The README does not provide a constrained trigger list, activation boundary, or negative examples to clarify when this skill should activate versus other shopping or assistant behaviors.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The trigger phrase "帮我买..." is broad enough to match ordinary conversational requests, which can cause the skill to activate in situations where the user did not explicitly intend to invoke a shopping workflow. In an agent setting with browser automation, accidental activation increases the risk of unintended web actions, navigation, or collecting product/account-adjacent context before the user realizes the skill is active.

Natural-Language Policy Violations

Medium

Confidence: 83% confidence
Finding: The workflow examples and interaction text assume Chinese-language input and output without first confirming the user's preferred language. This can lead to user misunderstanding of instructions, pricing, seller status, or handoff boundaries, which is especially risky in a shopping context where mistakes can affect purchases or trust decisions.

Natural-Language Policy Violations

Low

Confidence: 89% confidence
Finding: The example interaction is entirely in Chinese and models the skill as operating in that language, but the document does not state that Chinese is required nor does it offer an opt-in or alternative language. Under the language/locale policy, forcing a specific language without user choice can be a natural-language policy violation.

Static analysis

No suspicious patterns detected.