duckdb-analytics

name	duckdb-analytics
description	DuckDB in-process analytics tools for the CodeAgent — SQL on CSV/Parquet/JSON without a server

DuckDB Analytics — CodeAgent Tools

Overview

This skill extends the CodeAgent with DuckDB-powered data analytics functions. DuckDB runs in-process — no server required. Analyse CSV, Parquet, and JSON files with SQL, run aggregations, and export results.

Setup

pip install duckdb

CodeAgent Functions

• duckdb_query(sql, params=None) — run a SQL query, return results as list of dicts
• duckdb_sql(sql) — run a SQL query, return formatted table string
• duckdb_load_csv(path, table_name=None) — load a CSV file into a temp table
• duckdb_load_parquet(path, table_name=None) — load a Parquet file
• duckdb_load_json(path, table_name=None) — load a JSON/JSONL file
• duckdb_tables() — list all tables in the current session
• duckdb_describe(table) — describe a table's schema
• duckdb_export_csv(sql, path) — export query results to CSV

Example

# Load and analyse a CSV
duckdb_load_csv("sales.csv", "sales")
results = duckdb_sql("""
    SELECT region, SUM(amount) as total, COUNT(*) as orders
    FROM sales
    GROUP BY region
    ORDER BY total DESC
""")
print(results)

# Direct query on file — no load needed
result = duckdb_sql("SELECT * FROM 'data/*.parquet' LIMIT 10")
print(result)