Self-Hosted LLM Aider Configuration

This skill configures Aider to work with a self-hosted Ollama instance running Qwen2.5 32B, designed for local LLM development infrastructure on Mac Studio.

Configuration Overview

Create a `.aider.conf.yml` file in your project root with the following settings:

```yaml

Aider Configuration

This file contains Aider-specific configuration options

Model settings

model:

# Default model to use with Aider

name: "qwen2.5:32b"

# API base URL (uses docker network name)

api_base: "http://ollama:11434"

# Request timeout in seconds

timeout: 600

Editor settings

editor:

# Default editor to use

command: "vim"

# Whether to use dark mode

dark_mode: true

Git settings

git:

# Auto-commit changes

auto_commit: false

# Commit message prefix

commit_prefix: "feat: "

Chat settings

chat:

# Show model name in chat

show_model: true

# Show cost estimates

show_cost: false

# Enable streaming responses

stream: true

# Show diff before applying changes

show_diffs: true

File handling

files:

# Maximum file size to edit (in bytes)

max_file_size: 1048576

# File encoding

encoding: "utf-8"

# Auto-add files mentioned in chat

auto_add: true

Output settings

output:

# Format for code blocks

code_format: "markdown"

# Show line numbers in diffs

show_line_numbers: true

```

Key Features

Local Model Integration

Developer Experience

Git Workflow

Editor Integration

Setup Instructions

1. **Install Ollama**:

```bash

# For Mac

brew install ollama

```

2. **Pull Qwen2.5 Model**:

```bash

ollama pull qwen2.5:32b

```

3. **Start Ollama Service**:

```bash

# If using Docker Compose

docker-compose up -d ollama

# Or run locally

ollama serve

```

4. **Install Aider**:

```bash

pip install aider-chat

```

5. **Place Configuration**:

- Save the YAML configuration as `.aider.conf.yml` in your project root

- Aider will automatically detect and use it

6. **Verify Connection**:

```bash

aider --model qwen2.5:32b

```

Usage Examples

Start Aider Session

```bash

Use default configuration

aider

Explicitly specify model

aider --model qwen2.5:32b

Start with specific files

aider src/main.py src/utils.py

```

Common Commands

```bash

Add files to chat context

/add path/to/file.py

Review pending changes

/diff

Commit changes manually

/commit

Clear chat history

/clear

```

Docker Network Considerations

If running Ollama in Docker:

Customization

Adjust Model

```yaml

model:

name: "codellama:34b" # Or any Ollama-compatible model

```

Change Editor

```yaml

editor:

command: "code --wait" # VS Code

# or

command: "nano" # Nano

```

Enable Auto-commit

```yaml

git:

auto_commit: true

commit_prefix: "ai: "

```

Performance Tuning

Constraints