motion-figure
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill facilitates the creation of interactive figures by instructing the agent to generate ES5 JavaScript, HTML, and SVG content. This generated code is executed within the browser context via a provided runtime API (window.LF). This behavior is the intended primary function of the skill for educational visualizations and is subject to strict constraints defined in the SMIL and mechanism figure guidelines (SKILL.md, references/smil-rules.md).
- [INDIRECT_PROMPT_INJECTION]: The figure generation process relies on configuration data passed via JSON inside markdown fences. Ingestion points: JSON configuration strings within markdown figure fences (SKILL.md). Boundary markers: Absent for instructions within the data, but the input is parsed as JSON. Capability inventory: DOM and SVG manipulation, access to the window.LF API (references/toolkit.md), and WebGL rendering (templates/dither.js). Sanitization: Relies on the platform's JSON parser and the agent's adherence to generation rules. This creates a theoretical surface for untrusted data to influence the generated figure code, but no active exploitation patterns were detected.
Audit Metadata