Command reference · Aider

CLI Options (Flags)

Collects the most important Aider CLI flags, grouped by model, automation, context, and other settings.

Aider's CLI flags can be combined at startup (e.g. aider --model ... --test-cmd ... --yes-always) or set via a config file/environment variables. The most important ones, grouped: Model: --model MODEL (main model, env AIDER_MODEL); --weak-model WEAK_MODEL (for commit messages/history summarization); --editor-model EDITOR_MODEL (writes the file edits in architect mode); --architect (activates architect mode directly); --reasoning-effort VALUE and --thinking-tokens VALUE (control reasoning depth/budget) Automation/Scripting: --message/-m/--msg (send a single message, process the reply, then exit - for scripts/CI); --message-file/-f (read the message from a file); --yes-always (auto-confirms every prompt without asking); --no-auto-commits (disables automatic committing of LLM changes, default is on); --test-cmd (define a test command) together with --auto-test or a one-off --test; --lint-cmd (lint commands per language) together with --auto-lint or a one-off --lint; --watch-files (watches files for special AI comments that trigger Aider) Context: --read FILE (add a file read-only, usable multiple times) or --file FILE (add it editable); --map-tokens VALUE (repository map size, 0 disables it); --map-refresh VALUE (auto/always/files/manual) Install/Other: --upgrade/--update (update Aider itself); --just-check-update (only check whether an update is available); --config/-c (custom config file); --env-file (custom .env file for API keys); --version; --help/-h