skills/ceedaragents/cyrus/cyrus-setup/Gen Agent Trust Hub

cyrus-setup

Warn

Audited by Gen Agent Trust Hub on Mar 25, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: Uses shell scripts and background processes to identify open network ports and launch Google Chrome with remote debugging enabled to facilitate automation.
  • [DATA_EXFILTRATION]: Explicitly designed to extract authentication tokens, signing secrets, and API keys from web interfaces via automated scraping.
  • [REMOTE_CODE_EXECUTION]: Employs the agent-browser tool to perform eval operations, enabling the execution of arbitrary JavaScript within the context of a browser session.
  • [COMMAND_EXECUTION]: Instructs the agent to bypass standard file-handling tools in favor of direct shell manipulation (e.g., grep, printf) when interacting with sensitive environment files in ~/.cyrus/.
  • [DATA_EXFILTRATION]: Ingests untrusted data from external web platforms (Linear, Slack, GitHub) without sanitization or boundary markers, creating a surface for indirect instructions to influence agent behavior during setup.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 25, 2026, 10:20 PM