4 commands. 38 retailers. 8 countries. Your agent just learned how to shop.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
The problem
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
You're building an AI agent. It can search the web, reason about complex topics, generate code. But when it needs to answer a simple question like "how much does rice cost in Lima today?" — it hits a wall.
Not because it's dumb. Because every retailer has a different API, auth mechanism, schema, and checkout flow. There's no Stripe for commerce data.
Until now.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
4 lines. That's it.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
'''bash
# 1. Install
pip install cli-market-world
# 2. Guided first search + interactive tutorial
market init # login + first query
market tutorial # 3 exercises: search, compare, export — 60 seconds
# 3. Search — your agent just queried 38 retailers in <1s
market search "leche" --country PE --json
# 4. Connect to your AI agent via MCP
# Add to Cursor / Claude / any MCP client:
# → https://cli-market.dev/tools?utm_source=devto&utm_campaign=spike-jun09
'''
That's the entire integration. No API keys per retailer. No HTML scraping. No schema mapping.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
What happens under the hood
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
When your agent runs market search "rice" --country PE --json, CLI Market:
- Normalizes the query — "arroz", "rice", "arroz blanco" all map to the same product
- Queries our collector — 38 active retailers, refreshed every 4 hours
- Returns structured JSON — price per kg/L, store name, country, timestamp
- Handles auth — one CLI Market token, not 38 retailer tokens
What it looks like in your terminal:
![CLI Market terminal output — 60 seconds, real prices]
'''json
{
"query": "arroz",
"country": "PE",
"results": [
{
"product": "Arroz Extra 5kg",
"price": 18.90,
"price_per_kg": 3.78,
"store": "Wong",
"country": "PE",
"refreshed_at": "2026-06-10T08:00:00Z"
}
]
}
'''
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
MCP: the missing piece
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
CLI Market exposes 22 curated MCP tools (46 legacy) — search, compare, basket, checkout paths — so your agent can:
- Search products across all retailers
- Compare prices between stores
- Filter by country, category, or price range
- Get historical price data (Pro tier)
Connect in Cursor or any MCP client in 30 seconds. Configuration at cli-market.dev/tools.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
What you can build with this
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
- Price monitoring agents — track commodity prices across LATAM
- Shopping assistants — "find the cheapest 5kg rice in Lima"
- Competitive intelligence — benchmark your retail prices against the market
- Supply chain optimization — auto-reorder when prices drop below threshold
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
What's next
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
We're working on cart + checkout — so your agent doesn't just search, it buys. One API call to add to cart across any supported retailer.
Start building:
'''bash
pip install cli-market-world
market init
'''
**MCP tools:** [cli-market.dev/tools](https://cli-market.dev/tools?utm_source=devto&utm_campaign=spike-jun09)
**Docs for agents:** [cli-market.dev/llms.txt](https://cli-market.dev/llms.txt)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Pricing
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
| Tier | Price | Best for |
|------|-------|----------|
| Free | $0 | 1,000 req/day — prototype with `market init` |
| **Pro** | **$39/mo** | Alerts, full MCP tools, checkout, 20k req/day, 30d history — [activate](https://cli-market.dev//#pricing?utm_source=devto&utm_campaign=week2) |
CLI Market — Commerce infrastructure for AI agents. 38 retailers. 8 countries. MIT on PyPI.
Top comments (0)