Let Claude search Chefkoch recipes by rating and preparation time
To let Claude search Chefkoch, install Wield and name a dish or a few ingredients. For example: "Search Chefkoch for vegetarian lasagne recipes and list the best rated with preparation time and difficulty." Claude runs the chefkoch.de/search-recipes recipe in your own browser. It returns titles, ratings, votes, difficulty, preparation time and links. No login is needed.
Chefkoch Search cooking recipes with rating, difficulty and preparation time
Read-only, no loginPasses the daily check
Parameters (4)
| Metric | Wield recipe | Browser Use + Claude Sonnet 5.5 |
|---|---|---|
| Success | 100% | 100% |
| Median time | 414 ms | 40 s |
| Cost per run | $0 | $0.096 |
9 runs each, 7 November 2026. How we measure
What you need
- Claude Desktop, Claude Code, Cursor, Codex or another MCP client.
- A Chromium browser (Chrome, Edge, Brave, Arc, Vivaldi, Opera) with the Wield extension.
wield-mcp. Runnpx -y wield-mcp init, or use the one-click Claude Desktop bundle on wield.run/install.
For the full setup, follow the quick start.
Ask your agent
Name the dish or the ingredients. Add a sorting wish, such as "best rated" or "quick", and say how many results you want. Claude calls find_recipe to find the recipe, then run with query and any other parameters.
You: Search Chefkoch for vegetarian lasagne recipes and list the best rated with preparation time and difficulty.
Claude then runs chefkoch.de/search-recipes with query set to the dish. It reads the rating, votes, prep_time_min and difficulty from each result and sorts them for you.
Params: what to put where
| Param | Required | What to put in it |
|---|---|---|
query | yes | Dish or ingredients, for example lasagne, kürbissuppe or vegan schnell. |
limit | no | 10, 20 or 50 results. Default 20. |
page | no | Results page. Default 1. |
include_sponsored | no | true keeps brand-campaign recipes, which are marked sponsored. Default false. |
What comes back
The result has query, total, page and recipes. Each recipe shows an id, title, subtitle, rating, votes, difficulty, prep_time_min, a video flag, plus_only, sponsored, an image and the url.
difficulty is 1, 2 or 3, which means simpel, normal or pfiffig. plus_only marks recipes behind Chefkoch Plus, the paid tier. Recipe texts stay on chefkoch.de. Follow the link to read the method.
What leaves your computer
The recipe makes 1 request to api.chefkoch.de. The request goes out from your browser. Nothing passes through wield.run. No cookies or tokens are needed for this search.
Limits and terms
- Chefkoch's robots.txt disallows AI training crawlers (ClaudeBot, anthropic-ai, Claude-Web). Chefkoch's terms restrict automated use of its content. Wield runs the recipe only when you ask, one search at a time from your own browser. The scope is set: you start each search, one search per call, from your own browser, with no bulk collection and no purchases. Use it for your own cooking questions, not to copy recipes in bulk.
- Brand-campaign recipes are dropped by default. Set
include_sponsoredtotrueto keep them. - Status on 7 Nov 2026 is "ok". The check ran 5 of 5 checks in 1 request and 0.2 s.
In our benchmark (as of 7 Nov 2026, preliminary, still running), the recipe passed 9 of 9 runs with a median of 414 ms and no model cost. Browser Use with Claude Sonnet 5.5 passed 8 of 8 runs, with a median of 39.5 s and $0.10 per run. The check required at least five recipes with title and rating. The method is on the benchmark page.
FAQ
Does it need a login?
No. The search needs no login.
Can Claude read the full recipe?
The search returns the title, rating, difficulty, time and link. For the method and ingredients, open the link on chefkoch.de.
Is there an official Chefkoch 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. See AI apps that work with Wield.
How fast is it compared with a browser agent?
In our preliminary benchmark, the recipe's median was 414 ms. The browser agent's median was 39.5 s. Full details are on the benchmark page.