skills/zrong/skills/spritesheet/Gen Agent Trust Hub

spritesheet

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection via tool output poisoning. The generate_player function in scripts/spritesheet.py retrieves the video filename using video_path.stem and embeds it directly into the player.html template using string replacement (html.replace("{{TITLE}}", title)). Because the title is not HTML-escaped, a video file with a name containing script tags (e.g., \"><script>alert(1)</script>.mp4) will result in an player.html file that executes arbitrary JavaScript when opened in a browser.
  • Ingestion points: The video_path command-line argument processed in scripts/spritesheet.py.
  • Boundary markers: None present in the assets/player.html template.
  • Capability inventory: The skill performs local file writes (Path.write_text, cv2.imwrite) and executes local scripts via the uv run command.
  • Sanitization: Absent; the script performs raw string interpolation into an HTML template.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 01:20 PM
Security Audit — agent-trust-hub — spritesheet