bybit-trading

Warn

Audited by Gen Agent Trust Hub on Jul 17, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill implements a self-update mechanism and a bootstrap process that download and execute markdown modules and Node.js scripts from vendor-controlled GitHub repositories.
  • [COMMAND_EXECUTION]: The agent is instructed to run system commands such as curl for downloads, shasum for integrity checks, and node to execute the OAuth helper script. It also performs file system management operations like mkdir and mv.
  • [DATA_EXFILTRATION]: The skill manages sensitive API keys. While it uses them securely and implements redaction rules, the presence of financial credentials in the agent's context is a data exposure risk.
  • [PROMPT_INJECTION]: The skill uses background tasks for updates and includes priority rules that could override user instructions. It also implements defenses against indirect prompt injection from API data.
  • [EXTERNAL_DOWNLOADS]: The skill fetches code and instructions from api.bybit.com and raw.githubusercontent.com/bybit-exchange/. This allows for behavior modifications after the skill is initially loaded.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 17, 2026, 07:00 PM
Security Audit — agent-trust-hub — bybit-trading