concept-classifier
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- Prompt Injection (SAFE): No instructions designed to override agent behavior, bypass safety filters, or extract system prompts were detected in the skill description or metadata.
- Data Exposure & Exfiltration (SAFE): The skill does not access sensitive local files (e.g., SSH keys, environment variables) or hardcode any credentials. It describes a local data structure (
data.json) for quiz content. - Remote Code Execution & Dependencies (SAFE): While the skill mentions using p5.js (v1.11.10), it does not perform automated package installations (npm/pip) or execute remote scripts via curl/bash. The execution environment for the generated content is a web browser.
- Obfuscation (SAFE): The content is clear markdown without any Base64 encoding, zero-width characters, or homoglyph attacks.
- Persistence & Privilege Escalation (SAFE): The skill does not attempt to modify shell profiles, schedule tasks, or use
sudoto escalate permissions. - Indirect Prompt Injection (SAFE): Although the skill processes external quiz data via
data.json, this is used for rendering client-side UI and does not present a path for the data to influence the agent's internal control flow or execute dangerous capabilities.
Audit Metadata