Aider Multi-Model Configuration

This skill provides a comprehensive Aider configuration that enables professional AI-assisted software development with multiple model providers, automatic commits, and optimized settings for large codebases.

What This Skill Does

Configures Aider with:

Configuration

The skill sets up the following Aider configuration:

Model Selection Strategy

**Primary (Architect) Model:** `o3-mini` - Used for planning and complex reasoning

**Editor Model:** `anthropic/claude-3-7-sonnet-20250219` - Used for code implementation

**Weak Model:** `codestral/codestral-latest` - Used for simple refactoring and boilerplate

Core Features

1. **Architect Mode** (`architect: true`) - Plans changes before implementing

2. **Auto-commits** (`auto-commits: true`) - Automatically commits changes after successful edits

3. **Author Attribution** - Disabled to keep commits clean (`attribute-author: false`, `attribute-committer: false`)

4. **Cache Optimization** - Prompt caching enabled with 300s keepalive for cost efficiency

5. **Subtree Focus** (`subtree-only: true`) - Only shows relevant file tree sections

6. **Vim Integration** - Enabled for modal editing workflow

Token Management

Developer Experience

Usage Instructions

Step 1: Set Up API Keys

Before using this configuration, ensure you have API keys for your desired providers. Replace the placeholder values in the `api-key` section:

```yaml

api-key:

- openai=YOUR_OPENAI_KEY

- anthropic=YOUR_ANTHROPIC_KEY

- gemini=YOUR_GEMINI_KEY

- deepseek=YOUR_DEEPSEEK_KEY

- codestral=YOUR_CODESTRAL_KEY

- mistral=YOUR_MISTRAL_KEY

- github=YOUR_GITHUB_TOKEN

- cohere=YOUR_COHERE_KEY

- xai=YOUR_XAI_KEY

- groq=YOUR_GROQ_KEY

- together=YOUR_TOGETHER_KEY

- openrouter=YOUR_OPENROUTER_KEY

```

Remove or comment out providers you don't use.

Step 2: Save Configuration

Save this content to `.aider.conf.yml` in your project root or home directory (`~/.aider.conf.yml` for global config).

Step 3: Launch Aider

```bash

aider

```

Aider will automatically load the configuration and use the specified models.

Step 4: Leverage Architect Mode

When making complex changes, architect mode will:

1. Analyze the request

2. Create a plan

3. Show you the plan for approval

4. Implement the changes in stages

5. Auto-commit each successful change

Step 5: Customize for Your Workflow

**For testing workflows:**

Uncomment and configure:

```yaml

test-cmd: go test ./... -v -race -cover -timeout 10s

auto-test: true

```

**For custom system prompts:**

Uncomment and specify:

```yaml

read:

- SYSTEM_PROMPT_PROGRAMMING.xml

- PRELIMINARY_DESIGN.md

```

**For alternative API bases (local/custom endpoints):**

Uncomment and set:

```yaml

openai-api-base: http://localhost:8080/v1

```

Model Selection Guide

**Use o3-mini/o1** for:

**Use Claude 3.7 Sonnet** for:

**Use Codestral/Gemini Flash** for:

Important Notes

Constraints

Advanced Usage

Switching Models Mid-Session

In Aider prompt:

```

/model deepseek/deepseek-chat

/editor-model gemini/gemini-2.0-flash

```

Adding Files to Context

```

/add src/main.py

/add-folder src/components/

```

Manual Commit Control

If you prefer manual commits, set `auto-commits: false` and use `/commit` command when ready.

Reading Additional Context

Place context files (design docs, prompts) in your repo and reference them in the `read:` section. Aider will include them in every request.