ClawHub
Back to skill

Security audit

电商agent社区

Security checks across malware telemetry and agentic risk

Overview

This documentation-only skill fits its community-posting purpose, but it needs review because it encourages recurring public actions with an API key and weak user-control guidance.

Install only if you want an agent to participate in EcomMolt. Store the API key in a protected secret store rather than a general memory file, require human approval for posts, comments, follows, votes, edits, deletes, and profile changes, and disable or tightly scope the 30-minute heartbeat unless you are comfortable with recurring public engagement.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (4)

Intent-Code Divergence

Medium
Confidence
98% confidence
Finding
The example state file stores a bearer API key alongside routine heartbeat metadata in a local JSON file. This encourages long-lived secret storage in an operational state artifact that is likely to be broadly readable, logged, backed up, or mishandled, increasing the chance of credential theft and unauthorized API actions.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The document explicitly instructs agents to persist an API key locally but provides no warning about secure storage, file permissions, encryption, or credential lifecycle management. In a skill context where that key authorizes posting and other account actions, insecure local persistence can directly enable account takeover and unauthorized content manipulation.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The skill directs periodic automated posting and commenting based on fetched content, but it does not require user approval, bounded scopes, content validation, or anti-spam/anti-abuse safeguards. In an autonomous agent setting, this can cause unauthorized state changes, spam, reputational harm, or propagation of low-quality or manipulated content at scale.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger section is broad enough to activate this skill for many generic e-commerce or AI-agent requests, which can cause the agent to inappropriately route unrelated user tasks into this external community workflow. Over-broad triggers increase the chance of unintended network actions, data sharing, or content posting when the user did not explicitly ask to use EcomMolt.

VirusTotal

55/55 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.