Aider Integration Guide

Applicable roles: Developers, Admin Users, Member Users Last updated: 2026-08-06

Aider connects to the platform via the OpenAI Compatible adapter. The openai/ prefix in the Aider command is an adapter prefix — the model call name received by the platform does not include this prefix.

Installation

python -m pip install aider-install
aider-install

Quick Test with Environment Variables

macOS, Linux, or WSL:

export OPENAI_API_BASE="https://<your-api-endpoint>"
export OPENAI_API_KEY="YOUR_API_KEY"
aider --model openai/deepseek-v4-flash

Use the API endpoint shown on the console "API Key" page.

PowerShell:

$env:OPENAI_API_BASE="https://<your-api-endpoint>"
$env:OPENAI_API_KEY="YOUR_API_KEY"
aider --model openai/deepseek-v4-flash

After confirming the connection, you can write non-sensitive configuration to .aider.conf.yml:

openai-api-base: https://<your-api-endpoint>
model: openai/deepseek-v4-flash

It is better to keep the API key in environment variables. When you need to use a call name with a group identifier, the Aider parameter format is openai/<group-identifier>/<model-name>.

Verify in a Test Repository

First, ask a read-only question to confirm Aider can return results and usage records appear in the console. Then have it modify an easily verifiable piece of text and check with Git diff. This way, even if the model or tool behavior does not meet expectations, you can roll back directly.

If the model is not recognized, check whether the openai/ prefix is present. If the platform returns Model Not Found, check whether the model call name after the prefix matches the one from the console.