Security audit

my skill

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent interview-prep skill, but users should be mindful that resumes, job descriptions, files, URLs, and search queries may be processed to generate coaching content.

Install is reasonable if you want Chinese-language interview preparation. Before using it, remove unnecessary personal or confidential details from resumes and job descriptions, avoid private URLs unless needed, and only allow DOCX conversion or web searches when they clearly relate to materials you chose to provide.

SkillSpector

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
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The skill explicitly directs use of an `exec` capability to convert DOCX files via python-docx or pandoc. Allowing arbitrary command execution for document parsing is a dangerous overreach for an interview-prep skill because it expands the attack surface from content processing to code/tool execution, and malformed files or unsafe command composition could lead to command injection or processing of untrusted content in an unsafe runtime.

Context-Inappropriate Capability

Medium

Confidence: 81% confidence
Finding: The skill mandates external web searching for interview experiences before generating questions, which introduces network access and ingestion of untrusted third-party content beyond the core user-provided resume/JD workflow. This creates privacy and integrity risks: user job targets or company names may be sent to external services, and low-quality or manipulated web content could influence outputs without user awareness.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The trigger list includes broad natural-language phrases such as '帮我准备面试' and '出面试题', which can match ordinary conversation and cause the skill to activate unexpectedly. Over-broad invocation is risky because this skill can then process resumes, URLs, and web searches, leading to unintended data handling or tool use without the user deliberately selecting the skill.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill description does not disclose that it may process sensitive resumes, job descriptions, uploaded files, URLs, images, and externally searched web content. Because resumes commonly contain personal and employment data, the lack of transparency and consent around these inputs and tool calls increases the chance of privacy violations and user surprise.

Static analysis

No suspicious patterns detected.