Aider

Overview#

Aider is a command-line AI coding assistant that can edit code in your local git repository. It's designed for pair programming workflows and supports multiple LLM providers. Connect Aider to ModelStack to access all supported models from your terminal.

Prerequisites#

• Python 3.8+ installed
• Git repository initialized
• ModelStack API key from your dashboard

Installation#

Install Aider using pip:

python -m pip install aider-chat

Or using pipx (recommended):

pipx install aider-chat

Configuration#

Method 1: Environment Variables (Recommended)#

Set these environment variables in your shell:

export OPENAI_API_KEY="your_modelstack_api_key"
export OPENAI_API_BASE="https://api.modelstack.cc/v1"

Add these to your ~/.zshrc or ~/.bashrc to persist across sessions.

Method 2: Command-Line Flags#

Pass the configuration directly when running Aider:

aider \
  --openai-api-key your_modelstack_api_key \
  --openai-api-base https://api.modelstack.cc/v1 \
  --model claude-sonnet-4-6

Usage#

Basic Usage#

Start Aider in your project directory:

aider --model claude-sonnet-4-6

Aider will:

1. Detect files in your git repo
2. Connect to ModelStack
3. Open an interactive chat session

Example Session#

$ aider --model claude-sonnet-4-6

Aider v0.x.x
Model: claude-sonnet-4-6 via https://api.modelstack.cc/v1

> Add error handling to the login function in auth.py

# Aider will read auth.py, make changes, and show you a diff
# Type /yes to accept or /no to reject

Switching Models#

Change models mid-session:

> /model gpt-4o

Or specify at startup:

aider --model gemini-2.5-pro

Available Commands#

• /add <file> - Add files to the chat context
• /drop <file> - Remove files from context
• /model <name> - Switch to a different model
• /commit - Commit changes with AI-generated message
• /undo - Undo the last change
• /help - Show all commands

Troubleshooting#

Issue: "API key not found"

Solution: Ensure OPENAI_API_KEY is set or pass --openai-api-key flag.

Issue: "Model not found"

Solution: Use ModelStack's standardized model names. See supported models.

Issue: Rate limit errors

Solution: Implement exponential backoff and check your dashboard for account details if you are hitting rate limits frequently.

Best Practices#

1. Start small: Add only relevant files to context with /add
2. Use specific models: Choose faster models (Haiku, GPT-4o-mini) for simple tasks
3. Review changes: Always review diffs before accepting with /yes
4. Commit often: Use /commit to save progress incrementally

Next Steps#

• Explore model pricing to optimize costs
• Set up model stacks for automatic fallback
• Track token usage in your dashboard