website-to-hyperframes

Pass

Audited by Gen Agent Trust Hub on Apr 16, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted external data which creates a surface for indirect prompt injection.\n
  • Ingestion points: Untrusted data enters the agent context in references/step-1-capture.md via the hyperframes capture command, which extracts text, assets, and metadata from arbitrary user-provided URLs.\n
  • Boundary markers: The instructions do not define boundary markers or delimiters for the ingested text data processed in Step 1 and Step 3.\n
  • Capability inventory: The skill possesses significant capabilities, including shell command execution (npx hyperframes, node), file system writes for project files, and the use of external TTS APIs.\n
  • Sanitization: There is no evidence of sanitization or filtering of the captured website content before it is processed by the agent to generate summaries, scripts, and storyboards.\n- [COMMAND_EXECUTION]: The skill performs multiple shell command executions to manage its workflow.\n
  • Evidence: Frequent use of npx hyperframes for capturing, linting, validating, and rendering throughout the reference files.\n
  • Evidence: Execution of a local maintenance script in references/step-6-build.md via node skills/hyperframes-animation-map/scripts/animation-map.mjs.\n- [EXTERNAL_DOWNLOADS]: The skill references external libraries from established content delivery networks.\n
  • Evidence: references/techniques.md includes scripts from cdn.jsdelivr.net for Lottie and GSAP animations, which are well-known technology services.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 16, 2026, 07:04 PM