ClawHub
Back to skill

Security audit

resume-review

Security checks across malware telemetry and agentic risk

Overview

The skill does what it says, but it asks users to process full resumes with arbitrary third-party LLMs without a clear pre-send privacy warning.

Install only if you are comfortable having resume contents processed by your chosen AI tool or provider. Remove or mask names, contact details, IDs, school/company names, and other sensitive details before sending resume text to an external model, and be aware that extract_resume.py may write a raw resume text file locally.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Vague Triggers

Medium
Confidence
90% confidence
Finding
The README advertises very broad trigger phrasing such as telling an assistant to 'review my resume,' which can overlap with ordinary conversation and cause the skill to activate when the user did not explicitly intend to invoke it. In an agent environment, ambiguous activation increases the chance of unintended processing of sensitive resume data and unplanned execution of the skill's workflow.

Missing User Warnings

Medium
Confidence
86% confidence
Finding
The script encourages users to upload PDF or Word resumes for AI analysis and mentions anonymization, but it does not clearly warn that resumes contain sensitive personal data and that redaction may be incomplete or occur only after upload. In the context of an agent skill that processes personal documents, this omission can mislead users into sharing PII without understanding privacy, retention, or third-party model exposure risks.

Natural-Language Policy Violations

Medium
Confidence
92% confidence
Finding
The style guide hard-codes Chinese output and does not provide any mechanism to detect, respect, or ask for the user's preferred language. This can cause unwanted language switching, reduce accessibility for non-Chinese-speaking users, and create a poor or exclusionary user experience, especially if the agent is used in multilingual contexts.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The prompt explicitly instructs users to send full resume text to third-party LLMs, which commonly contains sensitive personal data such as names, phone numbers, email addresses, work history, and education details. Although the prompt later asks the model to anonymize output, that happens only after the data has already been disclosed to the external model provider, so the privacy risk remains.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.