---
name: agentic-storefront
title: Build an agentic auction storefront
description: Let an AI agent browse live auctions, monitor bids, and (with human confirmation) place bids or execute buy-now checkout on behalf of a buyer.
tags: [auctions, agentic-commerce, agents, buyers]
---

You are helping me build an agentic storefront that lets an AI agent participate in live auctions on behalf of a buyer.

**Buyer's preferences:**
- Max budget per item: [MAX_BUDGET_CENTS] cents
- Interested in categories: [CATEGORIES]
- Bidding strategy: [e.g. "bid just above current high, stop at max budget"]

Please do the following:

1. Call `auction_list_live` to see what auctions are currently running. Show me the results.

2. For each live auction that matches my buyer's interests:
   - Call `listing_get` to get full item details
   - Call `auction_get_high_bid` to get the current price
   - Evaluate whether it's within budget

3. For the best matching auction under budget, show me what `auction_place_bid` would look like. **Note: this requires my explicit confirmation before executing.** Present the bid details and ask for my approval before calling the tool.

4. If I approve, place the bid and show the result (success or outbid with new price).

5. Write a TypeScript agent loop (for a Node.js backend) that:
   - Polls `auction_list_live` every 30 seconds
   - Tracks auctions of interest
   - Fires a webhook or notification when an auction is ending in <5 minutes
   - Surfaces bid opportunities to the human buyer via a notification

6. Show me how to wire the agent loop with the Action Engine:
   - Register a signal `low_inventory` when auction ends in <5 minutes
   - Set a rule that sends a `ui_event: show_bid_opportunity` to the buyer's session

7. Show me how `listing_buy_now` works for fixed-price items and what the checkout URL flow looks like.