> ## Documentation Index
> Fetch the complete documentation index at: https://docs.2extract.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Price Monitoring
> Track prices across regions and marketplaces, with an agent that sets up the geo-targeted proxies for each market.
Prices differ by country, and marketplaces show you the ones for wherever they think you are. Checking a competitor's German pricing from a US IP gives you the US answer.
This is the case the MCP server handles best, because the hard part - working out which country codes exist and how to encode them - is exactly what the agent is good at.
***
## The setup
> Collect gaming-laptop prices from Amazon in Germany, the UK and Japan.
```text icon="terminal" theme={null}
listGeoCountries ⌐ DE · GB · JP
listProxyPlans ⌐ Pay-As-You-Go · residential
getAccountBalance ⌐ $149.95 available
createProxyResource ⌐ amazon_gaming_laptops
"Price tracking for gaming laptops on Amazon DE, GB and JP"
setProxyLimits ⌐ 50 GB per month
DE 2xt-customer-K3nP7qLm2Xv-proxy-amazon_gaming_laptops-country-de:s7Kd92mQ@proxy.2extract.net:5555
GB 2xt-customer-K3nP7qLm2Xv-proxy-amazon_gaming_laptops-country-gb:s7Kd92mQ@proxy.2extract.net:5555
JP 2xt-customer-K3nP7qLm2Xv-proxy-amazon_gaming_laptops-country-jp:s7Kd92mQ@proxy.2extract.net:5555
```
**One proxy, three markets.** The country lives in the username, so adding a fourth market later costs nothing - the agent just hands you another string.
***
## Wiring it into a scheduled job
```python monitor.py icon="python" theme={null}
import requests
BASE = "2xt-customer-K3nP7qLm2Xv-proxy-amazon_gaming_laptops"
PASSWORD = "s7Kd92mQ"
HOST = "proxy.2extract.net:5555"
MARKETS = {
"de": "https://www.amazon.de/s?k=gaming+laptop",
"gb": "https://www.amazon.co.uk/s?k=gaming+laptop",
"jp": "https://www.amazon.co.jp/s?k=gaming+laptop",
}
def fetch(country: str, url: str) -> str:
proxy = f"http://{BASE}-country-{country}:{PASSWORD}@{HOST}"
r = requests.get(url, proxies={"http": proxy, "https": proxy}, timeout=30)
r.raise_for_status()
return r.text
for country, url in MARKETS.items():
html = fetch(country, url)
# parse and store
```
Adding a market is one line in `MARKETS` - see the [country list](/data/countries) for available codes.
***
## Going below country level
National pricing is often the same everywhere; delivery estimates, stock, and local promotions are not. Narrow down when that matters:
> I need the same check from Munich specifically.
```text icon="terminal" theme={null}
searchGeoCities country_code: "DE", search: "Munich"
⌐ Munich · munich
...-proxy-amazon_gaming_laptops-country-de-city-munich:...
```
City targeting needs the country in the same username. Region and ZIP work the same way - see [Geo Targeting](/proxy-products/configuration/geo-targeting).
***
## Sessions: usually not what you want
Price checks are independent requests, so let each one take a fresh IP. That is the default with no `session` parameter, and it spreads the load naturally.
Use a session only for flows that break when the IP changes - adding to a cart to reveal a discounted price, or paginating through a result set that keeps server-side state. Then hold it briefly:
```
...-country-de-session-cart_check-time-5:...
```
***
## Keeping the bill flat
Monitoring runs on a schedule, so cost is a rate, not a one-off. Two habits keep it predictable.
**A cap that matches the schedule.** Four runs a day at \~120 MB each is roughly 15 GB a month. Set the ceiling a little above that:
> Cap the Amazon proxy at 20 GB a month.
If the site changes and pages get heavier, you hit the cap and the proxy stops - rather than quietly tripling the bill.
**A monthly review.** Ask, or run the `/monthly-cost-report` prompt:
> Show traffic and spend for the price-monitoring proxies last month.
Fetching only the search result page rather than every product page is usually the difference between 15 GB and 150 GB a month. Bandwidth is what you pay for.
***
## Next steps
Checking how a site renders in a given market.
Budgets, caps, and cleaning up what you stopped using.