Aider Configuration for Ollama DeepSeek-R1

Configure Aider to use Ollama's locally-hosted DeepSeek-R1 model with optimized settings for reliability and git integration.

What This Configuration Does

This Aider setup connects to a local Ollama instance running DeepSeek-R1, with safeguards against hanging requests, optimized retry logic, and git workflow customization. Ideal for developers running AI coding assistants entirely on local hardware.

Configuration Details

**Model Setup:**

**Git Integration:**

**Automation:**

Instructions

1. Prerequisites

Ensure you have Ollama installed and the DeepSeek-R1 model pulled:

```bash

Install Ollama (if not already installed)

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

Pull DeepSeek-R1 model

ollama pull deepseek-r1:latest

```

2. Create Configuration File

In your project root, create `.aider.conf.yml`:

```yaml

Project-Specific Aider Configuration

model: ollama/deepseek-r1:latest

auto-commits: false

yes-always: true

Ollama-specific settings to prevent hanging

env-file: .env

check-update: false

Timeout and retry settings (seconds)

timeout: 120

max-retries: 3

Model settings

model-settings:

temperature: 0.7

max-tokens: 4096

Prevent automatic git operations that might conflict

git: true

dirty-commits: true

attribute-author: false

attribute-committer: false

```

3. Optional: Create Environment File

If your project needs environment variables for Ollama (e.g., custom host):

```bash

.env

OLLAMA_HOST=http://localhost:11434

```

4. Launch Aider

```bash

aider

```

Aider will automatically detect and use the configuration file.

Configuration Options Explained

| Setting | Value | Purpose |

|---------|-------|---------|

| `model` | `ollama/deepseek-r1:latest` | Use DeepSeek-R1 via Ollama |

| `auto-commits` | `false` | Manual control over git commits |

| `yes-always` | `true` | Skip confirmation prompts |

| `timeout` | `120` | 2-minute timeout for model responses |

| `max-retries` | `3` | Retry failed requests up to 3 times |

| `temperature` | `0.7` | Balanced randomness in outputs |

| `max-tokens` | `4096` | Maximum response length |

| `dirty-commits` | `true` | Allow commits with uncommitted changes |

| `check-update` | `false` | Skip version check on startup |

When to Use This Configuration

Customization Tips

**Use a different Ollama model:**

```yaml

model: ollama/codellama:latest

```

**Increase timeout for slower hardware:**

```yaml

timeout: 300 # 5 minutes

```

**Enable auto-commits for faster workflow:**

```yaml

auto-commits: true

yes-always: false # Review commits interactively

```

**Adjust creativity level:**

```yaml

model-settings:

temperature: 0.3 # More deterministic

# or

temperature: 1.0 # More creative

```

Notes