/architect
Switches to architect mode, where one model plans the solution and a second model turns it into concrete file edits.
In architect mode, Aider first sends the request to the main model (the architect), which proposes a solution. Aider then sends that proposal to a separate editor model (configurable with --editor-model), which turns it into concrete file edits. Without an attached prompt, /architect only switches to this mode persistently, without sending a message.
โ WHEN TO USE IT?
Pair a strong reasoning model with a good editor model
/architect Refactor the payment module to use the new API โ e.g. with a strong reasoning model as architect and Sonnet or GPT-4o as the editor model
Give the same model two passes at the problem
/architect (no prompt, to switch persistently) โ the model gets one pass to plan and one to implement, which the docs say can produce better results than plain code mode
โ WHEN NOT TO?
A quick, simple change
Architect mode needs two LLM requests instead of one, which the docs note costs more time and money.
Better: Use /code for direct changes without the extra step
Just discussing, not changing anything
Architect mode ends up changing files; that's unnecessary if you only want to clarify questions.
Better: Use /ask to discuss the approach before any code gets changed
SOURCES
- Aider Docs โ In-chat commands โ aider.chat
- Aider Docs โ Chat modes โ aider.chat