ordercli

Warn

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill initiates the installation of a third-party binary via Homebrew (steipete/tap/ordercli) and Go (github.com/steipete/ordercli/cmd/ordercli@latest). These sources are outside the explicit trusted organizations list.
  • [CREDENTIALS_UNSAFE]: Instructions guide the agent to perform operations involving sensitive credentials, such as ordercli foodora login --email you@example.com --password-stdin and the use of the DELIVEROO_BEARER_TOKEN environment variable.
  • [COMMAND_EXECUTION]: The skill executes external CLI commands that access sensitive local file system paths, including browser profiles and application support directories.
  • [DATA_EXFILTRATION]: Commands like ordercli foodora cookies chrome and ordercli foodora session chrome explicitly target local browser cookie databases and session data, which are high-value targets for exfiltration.
  • [PROMPT_INJECTION]: The skill processes external data from Foodora/Deliveroo APIs (order history, restaurant names, and statuses). This represents an indirect prompt injection surface where malicious content in an order description could influence agent behavior.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 14, 2026, 12:13 AM