skills/simota/agent-skills/Realm/Gen Agent Trust Hub

Realm

Pass

Audited by Gen Agent Trust Hub on Mar 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The serve.py script executes local git commands (git log, git status) to gather activity data. These executions use hardcoded command lists and are performed through safe subprocess calls without shell expansion, preventing command injection.
  • [EXTERNAL_DOWNLOADS]: The visualization templates (templates/realm-game.html, templates/realm-map.html) download the Phaser game engine and typography from well-known technology providers (jsDelivr CDN and Google Fonts). These are documented as standard dependencies for the skill's visualization functionality.
  • [DATA_EXFILTRATION]: In live server mode, the skill starts an HTTP server that binds to all available network interfaces ('') by default on port 8765. This facilitates local network access to the visualization dashboard, which displays processed ecosystem data such as agent names and task summaries.
  • [PROMPT_INJECTION]: As the skill ingests untrusted data from git commit messages and agent journals for rendering in a browser-based dashboard, it includes a surface for indirect injection. The implementation mitigates this by applying HTML escaping to dynamic content before rendering in the UI.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 4, 2026, 03:08 AM