Aider Ollama Configuration

Configure Aider to use a local Ollama instance with the Qwen3 14B model, manual commit control, and Neovim editor integration.

What This Skill Does

This configuration sets up Aider to work with locally-hosted AI models via Ollama, giving you full control over your AI coding assistant while maintaining privacy and avoiding API costs. It configures Qwen3 14B as the model, disables automatic commits for better version control hygiene, and integrates with Neovim as the editor.

Instructions

1. **Verify Prerequisites**

- Ensure Ollama is installed and running locally

- Confirm the Qwen3 14B model is available: `ollama list`

- If not installed, pull it: `ollama pull qwen3:14b`

- Verify Aider is installed: `aider --version`

2. **Create the Configuration File**

- In your home directory or project root, create `.aider.conf.yml`

- Add the following configuration:

```yaml

##########################################################

Aider docs: https://aider.chat/docs/

##########################################################

model: ollama_chat/qwen3:14b

set-env:

- OLLAMA_API_BASE=http://127.0.0.1:11434

dark-mode: true

pretty: true

no-auto-commits: true

yes-always: false

editor: nvim

```

3. **Configuration Breakdown**

- `model: ollama_chat/qwen3:14b` — Use the Qwen3 14B model via Ollama's chat interface

- `set-env: OLLAMA_API_BASE` — Point to local Ollama server (default port 11434)

- `dark-mode: true` — Enable dark theme for terminal UI

- `pretty: true` — Enable syntax highlighting and formatted output

- `no-auto-commits: true` — Disable automatic git commits (you control when to commit)

- `yes-always: false` — Require confirmation for destructive operations

- `editor: nvim` — Use Neovim for file editing

4. **Test the Configuration**

- Navigate to a git repository

- Run: `aider`

- Verify it connects to Ollama and loads Qwen3 14B

- Test with a simple request: "Show me the main function"

5. **Usage Tips**

- **Manual commits**: After Aider makes changes, review them and commit manually with `git add` and `git commit`

- **Model switching**: Change the `model` line to use different Ollama models (e.g., `ollama_chat/codellama:34b`)

- **Editor preference**: Replace `nvim` with `vim`, `code`, or your preferred editor

- **Remote Ollama**: Change `OLLAMA_API_BASE` URL to connect to a remote Ollama instance

Customization Options

**Switch to a different Ollama model:**

```yaml

model: ollama_chat/codellama:34b # or deepseek-coder:33b, etc.

```

**Enable auto-commits (not recommended):**

```yaml

no-auto-commits: false

```

**Use a different editor:**

```yaml

editor: code # VS Code

or

editor: vim # Standard Vim

```

**Connect to remote Ollama:**

```yaml

set-env:

- OLLAMA_API_BASE=http://192.168.1.100:11434

```

Constraints

Why This Configuration