ClawHub
Back to skill

Security audit

私信通AB留资分析

Security checks across malware telemetry and agentic risk

Overview

The skill is a coherent AB-analysis playbook, but it handles private chat and contact-lead data and sends conversation content to a remote LLM without clear privacy, consent, or data-minimization controls.

Install only if the operator is authorized to process the private-message logs and to send sampled conversation text to the named remote LLM service. Before use, require redaction or pseudonymization of identifiers and contact details, keep outputs in a controlled directory, and confirm before updating the online report with --force.

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 (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding
The skill instructs writing analysis artifacts and reports to local paths and updating an online report, but no explicit permissions are declared. Undeclared write capability weakens reviewability and can lead to unintended file modification, data leakage into local workspaces, or abuse if the skill is invoked in a broader context than intended.

Vague Triggers

Medium
Confidence
83% confidence
Finding
The trigger phrases are broad and include generic requests like analyzing A/B differences, attribution, or rerunning reports, which can cause the skill to activate outside a narrowly scoped use case. Over-broad activation increases the chance the agent applies this workflow to unintended datasets or contexts, including sensitive conversation data, leading to inappropriate processing or report generation.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The document instructs operators to send session/conversation content, including session keys, group labels, industry, and compressed dialogue text, to a remote LLM endpoint, but it does not include any privacy notice, data minimization requirement, consent check, or handling constraints. Because the content explicitly involves private merchant chat sessions and lead-conversion analysis, transmitting raw conversational data to an external inference service creates meaningful confidentiality, compliance, and data-governance risk.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.