How to Seamlessly Add AI Coding Assistants to IntelliJ IDEA

Install AI Assistant plugins

Open Settings → Tools → AI Assistant → Agents . Click Browse repositories and install the desired AI coding‑assistant plugin (e.g., Claude Code, Codex). After installation open the AI Chat tool window and select the newly installed agent from the drop‑down list.

Connect third‑party AI providers

Navigate to Settings → Tools → AI Assistant → Providers & API keys . Available providers include:

Anthropic

OpenAI

OpenAI‑compatible API (custom endpoint)

Ollama (local models)

OpenAI‑compatible API configuration

Select OpenAI compatible API and fill in:

API URL – the base URL of the proxy or custom service (e.g., https://my-proxy.example.com/v1)

API Key – the authentication token for the service

This setup lets you route requests for multiple Codex accounts through a single endpoint, effectively creating an account‑pool proxy.

Local model integration via Ollama

If you run a model locally with Ollama, choose Ollama as the provider, then provide:

Model name – the identifier used by Ollama (e.g., llama2:7b)

API address – typically

http://localhost:11434

Note: When a Codex account is logged in, its credentials take precedence over the API configuration. Log out of Codex to force the IDE to use the custom API.

Using the AI Assistant

In the AI Chat window switch to Chat mode. Use the model selector in the lower‑right corner to pick the desired agent or model (e.g., Claude, Codex, a custom OpenAI‑compatible model, or an Ollama model). Then type coding questions or prompts as you would with any chat‑based code assistant.

Send a test prompt (e.g., “Write a Java method that reverses a string”) and verify that the response is syntactically correct and matches the expected behavior. Successful output confirms that the provider and model are correctly integrated.

Summary

IntelliJ IDEA’s built‑in AI Assistant provides a unified interface for:

Installing third‑party AI coding‑assistant plugins.

Connecting cloud providers (Anthropic, OpenAI) or custom OpenAI‑compatible endpoints.

Integrating locally hosted models via Ollama.

Switching between agents and models directly from the AI Chat panel.

This flexibility enables developers to use any supported large‑model API or local deployment without leaving the IDE.