ClawHub
Back to skill

Security audit

indeed-job-search

Security checks across malware telemetry and agentic risk

Overview

This skill is a purpose-aligned Indeed job extractor, but users should review its shell/Python helper usage before running batches.

Install only if you are comfortable letting the skill automate your browser on Indeed and use your active Indeed session for pagination. Review any generated bash batch script before running it, keep batches small, and avoid using it while unrelated sensitive account pages are open.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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)

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
The skill expands its operational scope from constrained browser actions to arbitrary shell execution via bash and local Python wrappers. That creates a much broader execution surface than the declared job-scraping function, and if the scripts or parameters are modified or tainted, the agent could execute unintended local commands or access resources outside the intended boundary.

Intent-Code Divergence

Medium
Confidence
91% confidence
Finding
The documentation claims no other tools are accessed, yet later instructs the agent to use bash to execute script wrappers. This inconsistency is dangerous because it can mislead reviewers and policy systems about the true capability set, reducing scrutiny of a pathway that enables broader code execution than advertised.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.