ordercli
Warn
Audited by Gen Agent Trust Hub on Apr 18, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill installs the
orderclitool from a third-party personal GitHub repository (github.com/steipete/ordercli) via Homebrew or Go. - [COMMAND_EXECUTION]: The skill triggers the execution of a binary that reads system-level browser profiles and session data from local application paths.
- [CREDENTIALS_UNSAFE]: Instructions guide the user to input sensitive authentication materials, including passwords through standard input, bearer tokens via environment variables, and importing cookies directly from web browsers.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface when processing external food delivery data. 1. Ingestion points: Order history and status details retrieved via
ordercli foodora historyin SKILL.md. 2. Boundary markers: No markers or instructions to ignore embedded instructions are present. 3. Capability inventory: The skill can perform state-changing actions such asordercli foodora reorderas defined in SKILL.md. 4. Sanitization: No data validation or sanitization of the fetched external content is identified.
Audit Metadata