ClawHub
Back to skill

Security audit

淘宝好券精选

Security checks across malware telemetry and agentic risk

Overview

This is a read-only Taobao coupon lookup skill that uses a disclosed cloud proxy and does not show evidence of purchases, account changes, local data collection, or persistence.

Install only if you are comfortable with coupon lookup requests being sent to the publisher's cloud proxy. The skill appears read-only and does not buy products or change accounts, but the embedded proxy token and limited backend disclosure mean you should treat the proxy service as part of the trust boundary.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding
The skill declares use of environment variables for a proxy service and clearly relies on network access, but no explicit permissions are declared. This creates a transparency and governance gap: reviewers and users cannot easily tell that external requests and secret-backed proxying are part of the skill’s operation, which can hide data flow to third parties or broaden the blast radius if the backend is changed later.

Vague Triggers

Medium
Confidence
80% confidence
Finding
The example trigger phrases are broad and conversational, increasing the chance that ordinary shopping-related user messages unintentionally invoke the skill. Accidental invocation can cause unsolicited external API calls, unexpected disclosure of user shopping intent to the service backend, and confusion about why the skill activated.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
This skill forwards user-supplied query parameters such as page, sort, and item_id to a remote proxy service without clearly disclosing that user inputs are transmitted off-box to a third party. In an agent-skill context, hidden exfiltration of user prompts or identifiers to external infrastructure is a genuine privacy and trust risk even if the data is not highly sensitive by design.

Env Variable Harvesting

High
Category
Data Exfiltration
Content
import urllib.error

PROXY_URL = os.environ.get("PROXY_URL", "https://1439498936-fzfcb544nx.ap-guangzhou.tencentscf.com")
PROXY_TOKEN = os.environ.get("PROXY_TOKEN", "tp_8k2mX9vQ4z")
TIMEOUT = 30
Confidence
97% confidence
Finding
os.environ.get("PROXY_TOKEN

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.