> ## Documentation Index > Fetch the complete documentation index at: https://docs.tryprofound.com/llms.txt > Use this file to discover all available pages before exploring further. # Introduction > End-to-end recipes for recreating the Profound dashboards with the public API. The Cookbook is a collection of **end-to-end recipes** that show how to recreate each chart, KPI, and table you see in the Profound app using only public API calls. Every recipe is self-contained: it lists the exact endpoints used, explains the request shape, and shows runnable Python and `curl` you can copy into your own code. If you just need the endpoint reference (every parameter, every metric), see the [REST API](/rest-api/introduction) tab. The Cookbook is the layer above that — common things people want to **build** with the API. ## Give your AI assistant context Building with Claude, ChatGPT, Cursor, or another AI coding assistant? Paste this URL — it's a compact index of every page in these docs that your assistant can fetch on demand: ``` https://docs.tryprofound.com/llms.txt ``` ## What you'll need Generate one in **Settings → API Keys** in the Profound app. See [Authentication](/rest-api/authentication) if you don't have one yet. Every report query is scoped to a category. See [Find your category ID](/cookbook/setup/find-your-category-id) to look one up by name. The recipes default to the Python SDK because it's the most common path. Install with `pip install profound`. Every recipe also has a `curl` tab if you'd rather hit the REST API directly. ## One-time setup Every recipe starts by constructing a client like this — you only need it once per script, even if you're chaining several recipes: ```python Python theme={null} import os from profound import Profound client = Profound(api_key=os.environ["PROFOUND_API_KEY"]) ``` ```bash curl theme={null} export PROFOUND_API_KEY=your_api_key_here ``` ## Conventions used in every recipe Three things to internalize before you start — these come up in every recipe and trip up most new users: The API parses `end_date` at the start of day, so it **excludes** the date you send. To include all of `2026-05-10`, send `end_date="2026-05-11"`. Display the inclusive value to your users; send `+1 day` to the API. Every response includes `info.query.metrics` and `info.query.dimensions`, which echo back the exact order the API used when packing each row's `metrics` and `dimensions` arrays. ```python theme={null} order = res.info.query["metrics"] # e.g. ["visibility_score"] i = order.index("visibility_score") score = res.data[0].metrics[i] ``` Don't hardcode positions. The order may not match the order you requested. Period-over-period changes (the green/red `+1.0 pp` you see on every KPI tile) are computed client-side. Run the same call twice — once for the current window, once for the previous window of equal length — and diff the two scores in your code. ## Before you build Rate limits, exclusive end-dates, `info.query` ordering, errors, pagination. Read this once, paste into your AI assistant. How Categories, Topics, Prompts, Tags, Assets, and Personas relate. Every endpoint, its metrics/dimensions, and what it's for — one scannable page. ## Recipes ### Setup List the categories your key can see and pick one programmatically. Get every asset in a category with its `is_owned` flag and domains. ### Visibility An asset's score for the window plus the change vs the prior window. Build the daily / weekly / monthly line chart for an asset. Fetch both at once. Understand why they're different calls. Rank every asset in a category by any visibility metric. Multi-line chart of any hand-picked set of assets. Break a single asset's score down by the AI surface that answered. ### Citations Owned-domain share of all citations, with the period-over-period change. Daily owned-domain share line **and** daily total citation volume, both from one call. Every cited domain ranked by share, with the per-row pp delta column.