Citation Chasing Mapping

• Outcome: Pass
• Latest audit: May 29, 2026, 8:23 PM
• Version:
  0.1.0

Overview

This is a coherent citation-network research helper with expected API lookups and local output files, not evidence of hidden or harmful behavior.

Install if you want a research tool that can run local Python, query Semantic Scholar, and write graph output files. Choose explicit output paths to avoid overwriting existing files, and avoid sending confidential unpublished research topics or private identifiers to third-party APIs unless that is acceptable for your workflow.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Missing User Warnings

Low

• Confidence: 78% confidence
• Finding: The examples write multiple local files such as network exports and visualizations without warning that existing files may be created or overwritten. In an agent setting, this can cause unintended modification of the user's workspace or loss of local data, especially when the model follows examples mechanically.

Vague Triggers

Medium

• Confidence: 85% confidence
• Finding: The skill summary is very broad and does not clearly constrain when the skill should be invoked, which can cause over-triggering in loosely related research tasks. In an agent setting, ambiguous activation increases the chance the skill is used on unintended inputs, potentially expanding exposure to prompt injection or causing unnecessary external data access and misleading outputs.

VirusTotal

64/64 vendors flagged this skill as clean.