seam-craft

Pass

Audited by Gen Agent Trust Hub on Jul 16, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a templating system for video transitions where tokens (e.g., __OLD__, __NEW__, __DX__) are substituted directly into JavaScript code.
  • Ingestion points: The gsap_template lines and the associated placeholder values are processed by an injector to be stamped into a script context.
  • Boundary markers: None identified; tokens appear to be direct string replacements within the JS templates.
  • Capability inventory: The system writes executable code to window.__timelines["main"] which is then seeked and rendered by the master clock.
  • Sanitization: The documentation explicitly states that the target environment (index.html) does not enforce the same whitelisting constraints (x/y/scale/rotation/opacity) as the scene-worker environment, and core/src/lint/rules/gsap.ts lacks a per-property whitelist for these master timeline injections.
  • [DYNAMIC_EXECUTION]: The skill describes the "stamping" of template code onto the master timeline at runtime. This involves generating and executing GSAP instructions dynamically based on the state of scene wrappers and track indices.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 16, 2026, 09:31 PM
Security Audit — agent-trust-hub — seam-craft