mouse-swarm-reveal

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires installing packages from the author's GitHub repository ('github:ruvnet/mouse-swarm') and uses various submodules under the '@ruvector' namespace for functionality like Ruflo integration and Playwright testing.
  • [COMMAND_EXECUTION]: Instructions provide local CLI commands for observation ('mouse-swarm observe'), page capture ('mouse-swarm-capture'), and video generation ('mouse-swarm-video'). These commands require local environment setup, including Playwright, Chromium, and FFmpeg for video processing.
  • [INDIRECT_PROMPT_INJECTION]: The skill has attack surfaces through the processing of local HTML files and agent status data. Ingestion points include the 'mouse-swarm-capture' command and the Ruflo adapter ('ruflo.js'), which processes status updates from a local bridge. The skill implements capability limits by focusing on rendering and provides sanitization in 'ruflo.js' and 'ruflo-view.js' to strip control characters and escape bracket characters in labels and task titles.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 04:07 PM