淘宝天天特卖

• Outcome: Pass
• Latest audit: Jul 15, 2026 · 2:37 AM UTC
• Version: 1.1.1

Overview

This skill is a Taobao deals lookup tool that uses a disclosed cloud proxy and shows no evidence of hidden persistence, local data access, purchases, or destructive behavior.

Install only if you are comfortable with Taobao shopping requests, product IDs, and sorting choices being sent through the skill publisher's cloud proxy. Avoid entering personal, account, payment, or private browsing information; this skill only provides product information and purchase links, not checkout or account actions.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Taint Tracking: Direct Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration

Findings (5)

Tainted flow: 'req' from os.environ.get (line 18, credential/environment) → urllib.request.urlopen (network output)

Critical

• Category: Data Flow

Content

headers={"Content-Type": "application/json", "X-Proxy-Token": PROXY_TOKEN},
    )
    try:
        with urllib.request.urlopen(req, timeout=TIMEOUT) as resp:
            return json.loads(resp.read().decode("utf-8"))
    except urllib.error.HTTPError as e:
        err = ""

• Confidence: 95% confidence
• Finding: with urllib.request.urlopen(req, timeout=TIMEOUT) as resp:

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 89% confidence
• Finding: The skill declares environment variables for a proxy endpoint and token and explicitly routes data through a networked proxy, but no corresponding permissions are declared. This creates a trust and review gap: operators and users cannot clearly see that the skill depends on outbound network access and secrets, which can hide data exfiltration or unsafe proxy behavior behind incomplete metadata.

Context-Inappropriate Capability

Medium

• Confidence: 86% confidence
• Finding: The skill is presented as a simple Taobao query tool, but it actually depends on a remote proxy endpoint and authentication token loaded from the environment. That hidden external dependency increases data-sharing and trust-boundary risk because user inputs are sent off-box to infrastructure not disclosed by the skill description.

Vague Triggers

Medium

• Confidence: 76% confidence
• Finding: The sample trigger phrases are broad everyday shopping utterances such as asking what good deals exist or requesting product details. Because these overlap with normal conversation, the skill may activate unintentionally and send user shopping context to the tool or backend without clear user intent, increasing privacy and routing risks.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: Tool parameters are serialized and sent to a remote proxy service without any user-facing notice or consent mechanism. Even though the parameters seem low sensitivity here, they can still include product IDs, browsing interests, and other query context that users may not expect to leave the local agent boundary.

VirusTotal

65/65 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.