microsim-vis-network
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEMETADATA_POISONINGINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [METADATA_POISONING]: The skill exhibits significant inconsistencies between its stated purpose and implementation details. The name and description specify the use of the "vis-network" library, but the technical implementation steps (Step 2) explicitly instruct the agent to use p5.js CDN links and JavaScript code. This discrepancy is deceptive and could lead to the generation of code that does not match the intended architectural scope.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a process where untrusted user input (Educational Requirements) is used to generate executable files.
- Ingestion points: User-provided subject area, learning objectives, and simulation concepts in the "Educational Requirements Specification" phase.
- Boundary markers: None identified to distinguish user-provided content from system instructions during the code generation process.
- Capability inventory: The skill instructs the agent to write multiple files (index.md, main.html, JS, and JSON) to the /docs/sims/ directory.
- Sanitization: No evidence of input validation, escaping, or sanitization before interpolating user requirements into generated executable scripts.
- [DYNAMIC_EXECUTION]: The skill's primary function is to generate and write JavaScript and HTML code at runtime based on dynamic user input. While central to the skill's purpose, this pattern inherently involves the creation of executable content from non-static sources.
Audit Metadata