doordash-group-orders

Pass

Audited by Gen Agent Trust Hub on Aug 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses dd-cli via shell commands to interact with DoorDash (searching restaurants, managing carts, and viewing order history). These are standard operations for a CLI-integration skill.
  • [INDIRECT_PROMPT_INJECTION]: The skill includes a 'Paste-a-thread intake' feature that parses chat logs from Slack or other platforms to build orders.
  • Ingestion points: Reads text pasted by the user representing chat threads in 'Flow 2'.
  • Boundary markers: None explicitly defined in the instructions for parsing the pasted content.
  • Capability inventory: The skill can execute dd-cli commands for cart management and write to local .dd/ JSON files.
  • Sanitization: The skill involves human review (Step 5: 'show the cart grouped by person') and human confirmation (Step 7) before finalizing orders, which mitigates the risk of an agent blindly following instructions embedded in a pasted chat thread.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 30, 2026, 12:54 AM
Security Audit — agent-trust-hub — doordash-group-orders