p5js

Pass

Audited by Gen Agent Trust Hub on Sep 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external data which could contain malicious instructions for the agent.
  • Ingestion points: Processes external datasets, API responses, and real-time inputs such as audio FFT data and user interaction events (referenced in references/interaction.md and SKILL.md).
  • Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore or isolate potentially malicious content within these data streams.
  • Capability inventory: The skill generates executable JavaScript code and uses local shell utilities (node, ffmpeg) for media processing (referenced in scripts/render.sh).
  • Sanitization: No evidence of input validation or sanitization for external data before it is interpolated into the generated p5.js code.
  • [DYNAMIC_EXECUTION]: The skill's primary function involves generating and executing JavaScript code to produce visual art.
  • Evidence: The agent generates self-contained HTML files with inline script tags containing p5.js logic and utilizes Puppeteer to execute these scripts for frame capture (referenced in scripts/export-frames.js).
  • [EXTERNAL_DOWNLOADS]: The skill references multiple external creative coding libraries via public Content Delivery Networks (CDNs).
  • Evidence: Fetches p5.js, p5.sound, and specialized addons from well-known services including Cloudflare (cdnjs), unpkg, and jsdelivr (referenced in templates/viewer.html and SKILL.md).
  • [COMMAND_EXECUTION]: The skill employs local command-line tools to automate the creative production pipeline.
  • Evidence: Utilizes node to run Puppeteer for headless browser sessions and ffmpeg to encode captured frame sequences into MP4 video files (referenced in scripts/render.sh and scripts/export-frames.js).
  • Evidence: The headless browser session is launched with the --disable-web-security and --allow-file-access-from-files flags to facilitate loading local file assets, which reduces the browser's security posture for that session (referenced in scripts/export-frames.js).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 13, 2026, 07:35 PM
Security Audit — agent-trust-hub — p5js