ClawHub
Back to skill

Security audit

腾讯会议面试助手

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed interview-assistance workflow that uses Tencent Meeting data or local files to prepare and evaluate interviews, with no evidence of hidden exfiltration, destructive behavior, or deceptive execution.

Install only if you are comfortable letting your agent use authorized Tencent Meeting contact and transcript capabilities for interview workflows. Confirm the candidate, meeting, resume, JD, and notes before running it, and treat generated scores as decision support rather than an automated hiring decision.

SkillSpector

Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Tp4

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding
The skill claims end-to-end Tencent Meeting integration and broad interview workflow automation, but the analysis indicates it only performs offline local file-based post-interview evaluation. This mismatch can mislead users and host agents into granting trust, providing sensitive candidate data, or invoking the skill in contexts where its actual controls, data sources, and limitations are not understood.

Vague Triggers

Medium
Confidence
78% confidence
Finding
The trigger phrases are broad enough to activate on common recruiting requests without clear scope boundaries, increasing the chance the agent runs on unintended inputs or sensitive HR conversations. In an interview-assistant context, overbroad activation is more dangerous because it may pull in or process candidate PII, evaluations, and meeting artifacts when the user did not intend to invoke this specific skill.

Static analysis

No suspicious patterns detected.