AI Coding Knowledge Base Configuration

This skill provides a production-ready Aider configuration optimized for AI-assisted development with automated testing, linting, and multi-model support.

What This Configuration Does

Key Features

1. **Model Aliases**:

- `sonnet` → Claude 3.5 Sonnet (main model)

- `gpt` → GPT-4o

- `o1` → OpenAI O1 Preview

- `gemini` → Gemini 2.0 Flash (editor model)

2. **Automated Linting** (runs after every change):

- `uv run black .` - Code formatting

- `uv run isort --profile black .` - Import sorting

- `uv run ruff check --fix` - Fast linting with auto-fixes

3. **Automated Testing**:

- Test command: `uv run pytest -v --disable-warnings tests/`

- Runs automatically after changes when `--auto-test` is enabled

4. **Quality of Life**:

- Auto-commits disabled (manual control)

- Yes-always mode for smoother flow

- Dark mode terminal colors

- Architect edit format for better code changes

Instructions for Use

1. Create Configuration File

Save this content as `.aider.conf.yml` in your project root or home directory:

```yaml

Main model configuration

model: sonnet

architect: true

editor-model: gemini

Model aliases for quick switching

alias:

- sonnet:claude-3-5-sonnet-20241022

- gpt:gpt-4o

- o1:openai/o1-preview-2024-09-12

- gemini:gemini/gemini-2.0-flash-exp

Automated quality checks

lint-cmd:

- uv run black .

- uv run isort --profile black .

- uv run ruff check --fix

auto-lint: true

Automated testing

test-cmd: uv run pytest -v --disable-warnings tests/

auto-test: true

Git and commit settings

auto-commits: false

yes-always: true

UI preferences

dark-mode: true

suggest-shell-commands: false

detect-urls: false

Read-only context files

read:

- CONVENTIONS.md

```

2. Adjust for Your Project

**Required changes**:

1. **Linting commands**: Replace with your project's linters

```yaml

lint-cmd:

- npm run lint --fix # For JavaScript/TypeScript

- cargo fmt # For Rust

- go fmt ./... # For Go

```

2. **Test command**: Update for your test framework

```yaml

test-cmd: npm test # JavaScript/TypeScript

test-cmd: cargo test # Rust

test-cmd: go test ./... # Go

```

3. **Read-only files**: Add project-specific context files

```yaml

read:

- CONVENTIONS.md

- ARCHITECTURE.md

- API_SPEC.md

```

3. Start Using Aider

```bash

Launch with configuration auto-loaded

aider

Override model for specific session

aider --model gpt

Use editor model alias

aider --editor-model gemini

```

4. Model Switching During Session

In the Aider chat:

```

/model gpt # Switch to GPT-4o

/model o1 # Switch to O1 Preview

/model sonnet # Back to Claude Sonnet

```

Configuration Highlights

Architect Mode

Enables more thoughtful, step-by-step code changes. Best for complex refactoring and feature implementation.

Dual-Model Strategy

Auto-Lint + Auto-Test

Catches issues immediately after changes. Combines well with `auto-commits: false` for manual review before committing.

Map Tokens

Not explicitly set, uses default 1024 tokens for repo context. Increase for larger codebases:

```yaml

map-tokens: 2048

```

Best Practices

1. **Start Simple**: Enable `auto-test` only after your test suite is reliable and fast

2. **Manual Commits**: Keep `auto-commits: false` for code review before committing

3. **Context Files**: Add architecture docs to `read:` for better AI understanding

4. **Model Selection**: Use `sonnet` for complex tasks, `gpt` for speed, `o1` for reasoning

5. **Lint First**: Ensure linters run cleanly before enabling `auto-lint`

Troubleshooting

**Linting fails**: Ensure dependencies are installed

```bash

uv pip install black isort ruff

```

**Tests timeout**: Disable auto-test or increase timeout

```yaml

auto-test: false

timeout: 120

```

**Model not found**: Verify API keys are set in `.env` file or environment variables

**Dark mode colors wrong**: Switch to light mode

```yaml

light-mode: true

dark-mode: false

```

Related Configuration Files