Back to feed
Dev.to
Dev.to
7/17/2026
AgentiQA: testing whether AI agents can complete real e-commerce checkout flows

AgentiQA: testing whether AI agents can complete real e-commerce checkout flows

Original: I sent an AI agent to buy from real stores. Here's what actually breaks.

Short summary

AgentiQA is an open-source (MIT) tool that sends a Claude-driven headless browser through real e-commerce checkout funnels to find where AI agents break. It pairs static checks (robots.txt, JSON-LD, sitemaps) with a live shopping agent that navigates product pages, adds to cart, and reaches checkout — with hard-coded payment field refusal in the executor layer. The tool supports both Anthropic API and Claude CLI subscription drivers to avoid per-run costs.

  • •Open-source AgentiQA tests if AI agents can actually complete purchases on real stores, not just lint markup
  • •Live shopper agent uses Claude with headless Chromium reading text-only DOM snapshots; payment fields are blocked at the executor level, not via prompt
  • •Supports both Anthropic API (~$1-2/audit) and Claude CLI subscription driver for zero per-run cost

Generated with AI, which can make mistakes.

#ai-tools#ai-agents#open-source
Read full article at Dev.to

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

AI Agent Checkout in Magento 2: Claude Places a Real Order via MCP
Dev.to
When the Buyer Is an AI: Optimizing Ecommerce for the Agent Shelf
Dev.to
Predictions first, data later: seven hot takes on AI agent readiness before I scan 50 sites.
Dev.to
The original headline is: "We ran an AI-agent-readiness scan on 19 AI-native brands. Here is what breaks."
Dev.to