Feynman Think Method

• Outcome: Pass
• Latest audit: Jun 1, 2026 at 3:43 PM UTC
• Version:
  2.0.1

Overview

This is a documentation-only learning-method skill with no hidden execution, data access, or persistence.

Safe to install for learning and explanation workflows. Be aware it may steer loosely related concept questions into a Feynman-method format and may answer primarily in Chinese unless the agent or user asks otherwise.

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)

Vague Triggers

Medium

• Confidence: 85% confidence
• Finding: The skill’s activation cues are very broad, everyday phrases such as 'I kind of understand but can’t explain it clearly,' which could cause the agent to invoke this skill in many loosely related contexts without clear scoping. Over-broad triggering is dangerous because it can override more appropriate task-specific behaviors, cause unsolicited pedagogical reframing, and increase the chance of the model applying the skill where the user did not intend it.

Natural-Language Policy Violations

Low

• Confidence: 80% confidence
• Finding: Specifying Chinese as the primary language without indicating user choice can cause the skill to respond in a language the user did not request, which is a prompt-behavior safety and usability issue. While not directly security-critical, it can lead to misunderstanding, consent issues around language preference, and unintended behavior when the surrounding system expects language preservation.

VirusTotal

64/64 vendors flagged this skill as clean.