Integrating Codex with Apache Airflow data via CData CLI

Codex is OpenAI's coding agent that can read, modify, and execute code on your machine, handling multi-file edits, running shell commands, reviewing code changes, and launching Codex Cloud tasks, all through natural language instructions. Its support for integrations, agent skills, and AGENTS.md configuration files makes it well-suited for structured, tool-driven workflows, making it a natural fit for connecting to external data sources through CData CLI.

By describing your data goals in plain language, Codex can handle the full setup process, from driver configuration and license activation to connection creation and query execution, without manual intervention at each step.

This article details step-by-step directions for how to connect Apache Airflow data to Codex through CData CLI.

Prerequisites

1. Codex installed (either Terminal CLI or Desktop)
   • Terminal CLI: CLI – Codex | OpenAI Developers
   • Desktop: App – Codex | OpenAI Developers
2. CData CLI installed
3. Access to Apache Airflow

Step 1: Download the skill (one-time setup)

Always use CData CLI with the official skill.

1. The official CData CLI Skill is available on GitHub and installs through npx skills in the terminal:

   npx skills add CDataSoftware/cli-skills

2. Follow the prompts in the terminal to install for Codex.

Step 2: Set up the project directory

Create a project directory to contain all project files. Below are examples for using Codex in either the Terminal or the Desktop app:

Codex in Terminal CLI

Navigate to the directory within the terminal and start a session with the codex command:

Codex Desktop App

Select the working directory within the desktop app:

Step 3: Establish the driver and connection

Describe what you want to accomplish in this session with the CLI and Apache Airflow data.

I would like to build a command line app that connects to Apache Airflow data and checks for updates from DagRuns. Make sure to include data from important columns like DagRunId and State.

This prompt automatically loads the skill and kicks off the following process. You can always manually prompt the agent for each of the following steps.

1. Driver setup: Codex checks for an existing CData Apache Airflow driver, or searches and downloads a new one:

• cdatacli drivers list

• cdatacli drivers search --driver "Apache Airflow"

• cdatacli drivers download --artifact-id <artifact-id>

2. Activation: Activate the Apache Airflow driver with a single command for a trial or full license:

• cdatacli drivers activate "Apache Airflow" --name "<name>" --email "<email>" --trial

• cdatacli drivers activate "Apache Airflow" --name "<name>" --email "<email>" --key "<product-key>"

3. Establish the Apache Airflow connection: Check for existing Apache Airflow connections or create a new one:

• cdatacli connection list

• cdatacli connection create --driver API --name my_api_connection --connectionstring "Prop1=value1;Prop2=value2;..."

4. Create a Apache Airflow skill (if applicable): CData provides driver instructions for popular sources. You can use these to generate a source-specific skill file that guides the agent through best practices for the driver.
• Run the following command and save the output to your skills directory, either at the project level or globally. (Note: If you receive a "No instructions available for Apache Airflow" message, no driver instructions exist for this source. You can continue using the main driver skill.)

  cdatacli drivers skill "Apache Airflow" > ~/skills/cdata-api/SKILL.md

Step 4: Query Apache Airflow data

With the CData driver fully configured, your agent can now execute queries and write code against live Apache Airflow data:

cdatacli query sql --connection my_api_connection --sql "SELECT * FROM DagRuns"

Query Apache Airflow data directly from your terminal with CData CLI

Codex and CData CLI together give your AI coding agent a direct path to live Apache Airflow data without custom middleware, scheduled syncs, or manual setup at each step. Describe your goal in plain language, and the agent handles driver configuration, connection setup, and query execution from start to finish in the terminal.

Download the free CData CLI and start a free, 30-day trial of the CData API Driver for JDBC today.