Multi-Model Aider Configuration
A comprehensive Aider configuration that enables seamless switching between multiple AI models, including free-tier options from various providers. Optimized for development workflows with proxy support and custom git integration.
Features
**Multiple model aliases**: Gemini, DeepSeek, Codestral, Cohere**Default models**: Gemini 2.5 Flash Lite (main), Gemini 2.0 Flash Lite (weak)**Proxy support**: HTTPS proxy configuration for restricted networks**Custom workflow**: Auto-commits disabled, custom gitignore patterns**Context file**: Reads AGENTS.md for additional contextConfiguration
This skill creates an `.aider.conf.yml` configuration file with the following setup:
Model Aliases
`main`: Primary model for coding tasks (Gemini 2.5 Flash Lite)`weak`: Lightweight model for simple tasks (Gemini 2.0 Flash Lite)`gemini`: Full-featured Gemini 2.5 Flash`deepseek`: DeepSeek Chat v3 (free tier via OpenRouter)`codestral`: Codestral-2501 via GitHub Models`cohere`: Command A 03-2025Network Configuration
HTTPS proxy: `127.0.0.1:18080`Google GenAI configured for direct API (not VertexAI)Environment variables loaded from `~/.secrets/envs`Git Integration
Git enabled but auto-commits disabled (manual commit control)Custom aiderignore using global gitignore patternsWatch-files disabled for better performancePerformance Optimizations
Cache prompts enabled for faster responsesModel warnings suppressedContext file (AGENTS.md) automatically includedUsage
1. **Initial Setup**: Create the configuration file in your home directory:
```bash
# Configuration will be saved to ~/.aider.conf.yml
```
2. **Environment File**: Create `~/.secrets/envs` with your API keys:
```
GEMINI_API_KEY=your_key_here
OPENROUTER_API_KEY=your_key_here
GITHUB_TOKEN=your_token_here
COHERE_API_KEY=your_key_here
```
3. **Switch Models**: Use aliases to switch between models:
```bash
aider --model gemini
aider --model deepseek
aider --model codestral
aider --model cohere
```
4. **Context File**: Create `AGENTS.md` in your project root for additional context that Aider should always consider.
Model Selection Guide
**Gemini 2.5 Flash Lite** (default): Best balance of speed and capability for most coding tasks**DeepSeek**: Excellent for complex reasoning, free tier via OpenRouter**Codestral**: Specialized for code generation via GitHub Models**Cohere**: Alternative model for diverse perspectivesFree Tier Resources
OpenRouter free models: https://openrouter.ai/models?max_price=0Google AI Studio: https://aistudio.google.com/app/prompts/new_chatGitHub Models: https://github.com/marketplace?type=modelsModel Rankings
OpenRouter rankings: https://openrouter.ai/rankings/programming?view=dayAider leaderboards: https://aider.chat/docs/leaderboards/Customization
To adapt this configuration:
1. **Change default models**: Uncomment desired model aliases in the config
2. **Add more aliases**: Define additional model aliases following the pattern
3. **Adjust proxy settings**: Modify or remove `HTTPS_PROXY` in `set-env`
4. **Custom gitignore**: Change `aiderignore` path to your preferred location
5. **Enable auto-commits**: Set `auto-commits: true` if desired
Notes
This configuration prioritizes free-tier models to minimize costsProxy configuration assumes local SOCKS5/HTTPS proxy on port 18080Environment file path assumes Linux home directory structureGit integration enabled but commits are manual for better control