Skip to main content
Codex CLI is OpenAI’s open-source local coding agent that runs in your terminal. You can connect it to NagaAI by defining a custom provider in ~/.codex/config.toml and using your NagaAI API key.

​
Setup

1

Install Codex CLI

Follow the official installation instructions: openai/codex
2

Create a NagaAI API key

Create an API key in the NagaAI Dashboard.
If you’re new to NagaAI, follow the Quickstart first.
3

Configure Codex CLI to use NagaAI

Open ~/.codex/config.toml and add a custom provider that points to NagaAI’s OpenAI-compatible API:
model_provider = "naga-ai"
model_reasoning_effort = "high"
model = "gpt-5.3-codex"

[model_providers.naga-ai]
name = "naga-ai"
base_url = "https://api.naga.ac/v1"
env_key = "NAGAAI_API_KEY"
gpt-5.3-codex is one example starter model. You can also use any NagaAI model that supports tools. See models with tools support.
4

Set your API key

Codex reads the API key from the environment variable named in env_key.
export NAGAAI_API_KEY="YOUR_NAGAAI_API_KEY"
Keep real API keys out of version control and shell history where possible.
5

Start Codex

In your project directory:
codex
Your requests will be sent through NagaAI using the provider defined in config.toml.

​
Notes

  • Config file: ~/.codex/config.toml
  • Base URL: https://api.naga.ac/v1
  • API key env var: NAGAAI_API_KEY
  • Example starter model: gpt-5.3-codex (browse tools-capable models)
  • If you use Codex project trust levels, configure them in the same config.toml file

​
Try it

  • “Summarize this repository: structure, build system, and entry points.”
  • “Find where feature X is implemented and list the relevant files.”
  • “Run tests and fix the first failing test.”
  • “Refactor module Y and update all call sites.”