idealo MCP server: let your AI agent look up current shop prices
Wield is an MCP server that gives your AI agent one idealo tool, idealo.de/price-lookup. The tool finds a product on idealo's price comparison and returns its shop offers, cheapest first. Each offer shows the shop, the item price, the price with shipping and the delivery estimate. The lookup runs from a tab in your own browser, and the requests go from your machine, not a cloud.
idealo Look up current shop prices for a product
Read-only, no loginRuns in your browser only
Parameters (2)
| Metric | Wield recipe | Browser Use + Claude Sonnet 5.5 |
|---|---|---|
| Success | 100% | 0% |
| Median time | 562 ms | 20 s |
| Cost per run | $0 | $0.025 |
9 runs each, 7 November 2026. How we measure
What your agent gets
| Recipe | What it does | Login | Status (7 Nov 2026) |
|---|---|---|---|
idealo.de/price-lookup | Looks up current shop offers for one product, with idealo's lowest and highest price and the offer count | No | OK, 5 of 5 checks in the browser |
The tool takes a query (a product name, such as "Sony WH-1000XM5") and max_offers, which is 20 by default and can be 40. The result names the product idealo actually used, so your agent can check the match. It also returns the product URL, the category, the lowest and highest price, the offer count and the offers.
In an in-browser check on 7 November 2026, the first product page returned 20 offer rows, with the lowest offer at 194,99 EUR.
How a request flows
- You ask your agent: "What is the cheapest price for the Sony WH-1000XM5 right now, including shipping?"
- The agent calls
find_recipe, thenrunwithidealo.de/price-lookupand the product name. wield-mcppasses the run through the local bridge to the Wield extension.- The extension makes two or three requests from an idealo.de tab, so they carry your normal browser session.
- The offers go back to your agent. Nothing passes through wield.run.
idealo's bot protection rejects requests from servers. This recipe therefore runs only in your browser.
What leaves your computer
The only host is www.idealo.de. A lookup makes two or three requests. The recipe never follows idealo's tracked links to shops, so a lookup triggers no click-outs.
Wield logs each request on your computer only: method, host, path, status and timing. It never logs bodies or query values. Telemetry is off by default. Wield never reads, copies or sends your cookies or passwords.
idealo terms, in plain words
We have not reviewed idealo's terms as a lawyer would, so this is not legal advice. Our reading is that idealo's terms prohibit automated extraction. This recipe runs one lookup that you start, from your own browser, at human pace. It does not loop over products and does not click through to shops. Check idealo's terms yourself before you automate anything else.
Limits and what is not working yet
- Read only. The tool compares prices. It does not order anything.
- The first suggestion wins. The tool uses idealo's first product suggestion for the query. Check
productin the result before you trust the prices. - Browser only. idealo blocks server requests, so the tool works only where your own browser runs.
- Prices are a snapshot. They are what idealo shows at the moment of the lookup.
- Run limits apply. A run has at most 12 steps, 40 requests, 5 MB per response and 60 seconds.
FAQ
Is there an official idealo MCP server?
Not that we know of as of 7 November 2026.
Which AI apps work with it?
Any MCP client. npx -y wield-mcp init sets up Claude Desktop, Claude Code, Cursor and Codex. Other MCP clients work with the standard stdio config.
Does it need an idealo login?
No. Price lookups run without an account.
How fast is it compared with a browser agent?
The benchmark task checks for at least five shop offers with a shop name and price. As of 7 November 2026, and still preliminary, the Wield recipe passed 9 of 9 runs with a median of 562 ms and no model cost. Browser Use with Claude Sonnet 5.5 passed 0 of 9. Bot protection blocked all 9 of its runs. The median of 20.1 seconds and $0.03 per run describes those blocked runs, not a time to a result. See the idealo benchmark.