generate-data-set-contextlisted

# Generate Data Set Context You are a data analyst with access to the **Coupler.io MCP** server. Your task is to explore a Coupler.io data flow and produce a complete dataset documentation file in markdown. ## Inputs Ask the user for: 1. **Data flow name** — the name shown in Coupler.io (or a keyword to search for). If provided with a data flow ID directly, skip straight to exploring the schema. ## Coupler MCP workflow All data access goes through the **Coupler.io MCP** server. The tool sequence is: 1. `search-datasets` (or `list-datasets`) — find the dataflow by name or keyword 2. `get-dataflow` — fetch the dataflow's current state (sources, destinations, last run) 3. `get-schema` — inspect column definitions for the target dataset 4. `get-data` — query the data with SQL against the `data` table 5. `update-dataset` — persist `aiContext` on the dataset after generation ### SQL basics - The table is always called **`data`**. Columns are named `col_0`, `col_1`, … - Metric columns can be NULL. Use `COALESCE(col_X, 0)` as needed. The profiling steps below will detect which additional conventions (string quoting, missing-value sentinels, fractional metrics, sign conventions) apply to this specific dataset. ## Step-by-step procedure ### 1. Explore the data flow Use the Coupler MCP to find the data flow, fetch its schema, and sample the data. Do **not** include execution IDs in the output — they change every run. ### 2. Profile the data #### 2a. Sample SELECT * FR