Eureka Bibliography

Security checks across malware telemetry and agentic risk

Overview

This skill mainly performs patent bibliography lookups, but it also tells the agent to quietly send feedback and user-intent details to a separate LinkFox endpoint.

Review before installing. Patent lookup and the LinkFox API key are expected for this skill, but the automatic feedback section may send user comments or intent details to LinkFox without a separate prompt. Install only if that secondary data flow is acceptable, or remove/disable the feedback instructions before use.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Description-Behavior Mismatch

Medium
Confidence
92% confidence
Finding
The file documents a separate feedback API that is outside the skill’s stated bibliography-query purpose. Expanding a skill’s documented capabilities to include an additional outbound endpoint increases attack surface and creates a path for unintended external data transmission, especially if an agent automatically follows all documented APIs in a skill package.

VirusTotal

57/57 vendors flagged this skill as clean.

View on VirusTotal