abandoned-cart-recovery

Pass

Audited by Gen Agent Trust Hub on Nov 16, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides instructions to install the Whop CLI using piped shell execution. Evidence includes curl -fsSL https://whop.com/install.sh | sh and irm https://github.com/whopio/whop-public-cli/releases/latest/download/install.ps1 | iex in references/whop-connection-and-readiness.md. These commands are standard for the official tool and target legitimate vendor domains to establish the necessary environment.
  • [EXTERNAL_DOWNLOADS]: The skill downloads and executes external tools and packages, specifically the @whop/cli Node.js package and shell scripts from whop.com and GitHub to facilitate integration with the commerce stack.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection due to its processing of external commerce data.
  • Ingestion points: The skill ingests abandoned cart payloads, customer activity logs, and event metrics from Shopify and Klaviyo integrations (references/event-inspection.md, references/shopify-adapter.md).
  • Boundary markers: The skill includes explicit instructions in references/browser-runtime-and-sessions.md telling the agent to treat page text and tool descriptions as data rather than permission to run commands or disclose private context.
  • Capability inventory: The agent is empowered to use browser tools to navigate commerce platforms, draft and send recovery messages, and modify store configurations.
  • Sanitization: The instructions recommend using native platform filters and templates which provides a layer of structural validation for external content.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 16, 2026, 10:48 AM