Init Duckdb Workspace

Workflow

1. Scaffold Directories: Create the following structure in the project root:

   • duckdb/data/csv/
   • duckdb/data/parquet/
   • duckdb/data/json/
   • duckdb/db/public/
   • duckdb/query_editor/_template/

2. Deploy README: Copy the standardized assets/README.md to duckdb/README.md.

3. Deploy Guide: Create duckdb/query_editor/_template/guide.sql with the following content:

   -- 1. Query files directly (In-memory)
   SELECT * FROM read_csv_auto('duckdb/data/csv/*.csv') LIMIT 10;
   SELECT * FROM read_parquet('duckdb/data/parquet/*.parquet') LIMIT 10;
   SELECT * FROM read_json_auto('duckdb/data/json/*.json') LIMIT 10;
   
   -- 2. Create/Open persistent DB
   -- Command: duckdb duckdb/db/public/my_data.db
   
   -- 3. Create tables in persistent DB
   CREATE TABLE my_table AS SELECT * FROM read_csv_auto('duckdb/data/csv/my_file.csv');
   
   -- 4. List existing tables
   SHOW TABLES;
   

Instructions for the Agent

• Use the assets/README.md file provided in this skill as the source of truth for the workspace documentation.
• Ensure all directory paths are relative to the workspace root.
• If the duckdb/ directory already exists, ask the user for confirmation before overwriting or appending.