About

This post assumes that you already have a preferred Emacs configuration and you’re here to learn about integrating aider into your development workflow.

From their website:

aider lets you pair program with LLMs to start a new project or build on your existing codebase.

Setup

First, install aider by executing:

python -m pip install aider-install
aider-install

Aider Standalone

Before you integrate aider with your emacs configuration you can test the agent from your terminal. Taken from the main wesbite:

# Change directory into your codebase
cd /to/your/project

# DeepSeek
aider --model deepseek --api-key deepseek=<key>

# Claude 3.7 Sonnet
aider --model sonnet --api-key anthropic=<key>

# o3-mini
aider --model o3-mini --api-key openai=<key>

Install Emacs Packages

Make sure that you install aider locally like in the step above:

Verify that your Emacs version is 30 or greater. Execute the following to upgrade your emacs installation:

sudo add-apt-repository ppa:ubuntuhandbook1/emacs
sudo apt-get update
sudo apt-get install emacs

After executing emacs --version you’ll see that you have the latest version installed.

You may need to re-install some packages and libraries to match your desired configuration.

Open Emacs and install the aider and aidermacs packages via M-x list-packages. Find the desired packages, press i next to each package, then x to install them.

Add the following to your init.el file:

(use-package aidermacs
  :bind (("C-c a" . aidermacs-transient-menu))
  :config
  ; Set API_KEY in .bashrc, that will automatically picked up by aider or in elisp
  ;(setenv "ANTHROPIC_API_KEY" "sk-...")
  ; defun my-get-openrouter-api-key yourself elsewhere for security reasons
  ;(setenv "OPENROUTER_API_KEY" (my-get-openrouter-api-key))

  (setq aidermacs-default-model "ollama/deepseek-coder-v2")
  )

Once you restart Emacs you should now be able to access the aider agent with the key binding C-c a. See the Usage section for more details on how to interact with aidermacs.

Models

One of the beautiful things of aider is that you can choose whatever AI Model you want (Gemini, DeepSeek, OpenAI, etc). There are multiple of free models you can use as well. This is a list of available models to choose from:

https://aider.chat/docs/llms.html

For example, to use the openrouter/google/gemini-2.0-flash-exp:free model, update your init.el file such that the aidermacs-default-model field points at the model you want to use.

Run Models Locally

aider can connect to local Ollama models. [Referece].

First install Ollama. In a Linux terminal, execute the following:

curl -fsSL https://ollama.com/install.sh | sh

You’ll see an installation message:

Created symlink /etc/systemd/system/default.target.wants/ollama.service → /etc/systemd/system/ollama.service

You should now be able to check on the status of ollama by executing:

systemctl status ollama

Start Ollama (or your local inference server) and ensure your model is pulled. For example:

ollama pull deepseek-coder-v2

Run aider with the --model flag pointing to the local endpoint:

export OLLAMA_API_BASE=http://127.0.0.1:11434
aider --model ollama/deepseek-coder-v2

Note:

When running models locally, consider that they’ll run slower depending on your machine’s GPU.

Usage

In our init.el we’ve defined the binding C-c a to access the aidermacs-transient-menu (Figure 1). This menu summarizes the tools available to you and the corresponding keys to execute/access each one.

I won’t summarize each tool (mainly becuase I’m still learning myself), but the best way to learn is to dive in and play!

Scripting Aider

It is possible to write Python scripts to interact with aider. I have not yet personally played with this functionality, but you can read more here.