Aider Configuration for Statistical Business Register

This skill provides a production-tested Aider configuration optimized for the Statistical Business Register (statbus) project, particularly for managing large SQL codebases with quality control workflows.

Key Features

Configuration Details

Automatic Context Loading

The configuration automatically reads key project files to provide context:

```yaml

read: [CONVENTIONS.md, app/CONVENTIONS.md, doc/data-model.md]

```

This ensures Aider understands:

Working Files

Editable files for ongoing work:

```yaml

file: [todo.md, tmp/journal.md]

```

These track tasks and development notes. The `add-gitignore-files: true` setting allows editing these temporary files even when gitignored.

Quality Control Settings

```yaml

auto-commits: false

auto-lint: false

```

**Why auto-commits are disabled:** Large SQL file edits require manual review before committing. This prevents unreviewed code from entering the codebase.

**Why auto-lint is disabled:** SQL files don't pass default linting rules. Custom lint commands can be configured when needed.

File Watching

```yaml

watch-files: true

```

Monitors files for AI coding comments, enabling reactive development workflows where comments trigger Aider actions.

Repository Scope

```yaml

subtree-only: true

```

Restricts Aider operations to the current subtree of the git repository, preventing unintended changes to other project areas.

How to Use

1. **Copy this configuration** to `.aider.conf.yml` in your statbus repository root

2. **Adjust context files** to match your project structure:

```yaml

read: [YOUR_CONVENTIONS.md, YOUR_DATA_MODEL.md]

```

3. **Set working files** for your workflow:

```yaml

file: [your-todo.md, your-notes.md]

```

4. **Launch Aider** - it will automatically load the configuration and context files

5. **Make changes** - Aider will assist with edits but won't auto-commit

6. **Review diffs** manually before committing (use `git diff` or `/diff` command in Aider)

7. **Commit when ready** using Aider's `/commit` command or standard git workflow

Workflow for SQL Development

1. Start Aider with context already loaded from conventions and data model docs

2. Work on SQL files with Aider's assistance

3. Review all changes carefully (especially large SQL edits)

4. Manually approve and commit changes

5. Use `tmp/journal.md` to track development progress

Customization

Enable Auto-Linting for SQL

If you have SQL-specific linters:

```yaml

auto-lint: true

lint-cmd: ["sql: sqlfluff lint --dialect postgres"]

```

Add More Context Files

```yaml

read: [CONVENTIONS.md, app/CONVENTIONS.md, doc/data-model.md, doc/api-spec.md, ARCHITECTURE.md]

```

Enable Auto-Commits for Non-SQL Work

For Python/TypeScript development on the same project:

```yaml

auto-commits: true

```

Or use separate `.aider.conf.yml` files in different subtrees.

Important Notes

Related Configuration

For advanced customization, refer to the full Aider configuration options in the source comments, including:

Best Practices

1. **Keep conventions up to date** - Aider reads them automatically, so they should reflect current practices

2. **Use journal files** - Track complex changes in `tmp/journal.md` for future reference

3. **Review before committing** - The disabled auto-commit is a feature, not a limitation

4. **Leverage file watching** - Use AI coding comments in source files for context-aware assistance

5. **Scope appropriately** - Use `subtree-only: true` for large, multi-team repositories