DevTrack Aider Configuration

A comprehensive Aider configuration skill that demonstrates all available settings for customizing the AI pair programming assistant. This configuration covers model selection, API keys, git integration, linting, testing, output customization, and workflow preferences.

Instructions

You are configuring Aider, an AI pair programming tool, for a development project. Follow these steps to create a complete `.aider.conf.yml` configuration file tailored to the project's needs.

Step 1: Determine Model Configuration

Ask the user which AI model they want to use for the main chat:

Determine if they want a separate weak model for commit messages and summarization (usually cheaper/faster model).

Step 2: Configure API Keys and Base URLs

Determine where API keys should be stored:

If using custom API endpoints, ask for:

Step 3: Configure Git Integration

Determine git workflow preferences:

Ask if they want to:

Step 4: Configure Code Quality Tools

**Linting:**

**Testing:**

Step 5: Configure Output and UI Settings

Determine display preferences:

Step 6: Configure Repo Map and Context

Set context window preferences:

Step 7: Configure History Files

Customize history file locations:

Step 8: Configure Advanced Options

Determine additional preferences:

Step 9: Generate Configuration File

Create a `.aider.conf.yml` file with the selected options. Structure it in logical sections:

```yaml

##########################################################

Aider Configuration for [Project Name]

##########################################################

#######

Main:

Model selection

model: [selected-model]

weak-model: [selected-weak-model] # Optional

API Configuration

openai-api-key: [key] # Only if not using .env

anthropic-api-key: [key] # Only if not using .env

#################

Model Settings:

map-tokens: 1024

cache-prompts: false

edit-format: [format] # If specific format required

###############

Git Settings:

auto-commits: true

dirty-commits: true

attribute-author: true

attribute-committer: true

commit-prompt: [custom prompt] # Optional

########################

Fixing and committing:

auto-lint: true

lint-cmd:

- "python: flake8 --select=E,W,F"

- "javascript: eslint"

test-cmd: [test command]

auto-test: false

##################

Output Settings:

dark-mode: true # or light-mode: true

pretty: true

stream: true

code-theme: default

################

History Files:

Defaults are usually fine, customize if needed

input-history-file: .aider.input.history

chat-history-file: .aider.chat.history.md

#################

Other Settings:

suggest-shell-commands: true

check-update: true

vim: false # Set true for VI editing mode

```

Step 10: Create .aiderignore File (Optional)

If the user wants to exclude certain files or directories from Aider's context, create a `.aiderignore` file with gitignore-style patterns:

```

Dependencies

node_modules/

venv/

__pycache__/

Build outputs

dist/

build/

*.pyc

Sensitive files

.env

*.key

*.pem

Large data files

*.csv

*.json # If data files

datasets/

```

Step 11: Document Usage

Provide usage instructions for the team:

**Basic usage:**

```bash

Start aider with config file (auto-detected in git root)

aider

Edit specific files

aider file1.py file2.js

Add read-only context files

aider --read docs.md --file main.py

Use a different model for this session

aider --model gpt-4o

Run in lint-and-fix mode

aider --lint

Run in test-and-fix mode

aider --test

```

**Configuration override:**

```bash

Use a different config file

aider --config .aider.custom.yml

Disable auto-commits for this session

aider --no-auto-commits

Show repo map and exit (debugging)

aider --show-repo-map

```

Step 12: Provide Best Practices

Share these recommendations with the user:

1. **API Keys**: Store in `.env` file, never commit to git

2. **Model Selection**: Use Sonnet for most tasks, Opus for complex reasoning, Mini for speed

3. **Auto-commits**: Keep enabled for better history tracking

4. **Linting**: Configure lint commands to match project's CI/CD pipeline

5. **Testing**: Enable auto-test only if tests are fast (<30s)

6. **Repo Map**: Increase `map-tokens` for large codebases (2048-4096)

7. **Prompt Caching**: Enable for models that support it (saves costs on long sessions)

8. **Chat History**: Review `.aider.chat.history.md` periodically for learnings

9. **Subtree Mode**: Use `--subtree-only` when working on a specific module

10. **Read-only Files**: Use `--read` for context files you don't want modified (docs, specs)

Examples

Example 1: Python Data Science Project

```yaml

model: claude-3-5-sonnet-20240620

weak-model: claude-3-haiku-20240307

auto-lint: true

lint-cmd:

- "python: black --check ."

- "python: mypy --strict"

test-cmd: "pytest --cov=src tests/"

auto-test: false

map-tokens: 2048

dark-mode: true

```

Example 2: JavaScript Full-Stack Project

```yaml

model: gpt-4o-2024-08-06

weak-model: gpt-4o-mini

auto-lint: true

lint-cmd:

- "javascript: eslint --fix"

- "typescript: tsc --noEmit"

test-cmd: "npm test"

auto-test: false

map-tokens: 1024

code-theme: monokai

```

Example 3: Enterprise Project with Custom API

```yaml

model: azure/gpt-4

openai-api-base: "https://custom.openai.azure.com/"

openai-api-type: "azure"

openai-api-version: "2024-02-15-preview"

openai-api-deployment-id: "gpt-4-deployment"

auto-commits: true

attribute-commit-message-committer: true

map-tokens: 4096

verify-ssl: true

```

Constraints

Notes