CodeRabbit AI Automation

Process CodeRabbit AI code review comments from GitHub PRs and automatically apply suggested fixes to codebases using a Python-based automation toolkit.

What This Skill Does

This skill provides guidance for working with a Python-based automation toolkit that:

Architecture

The toolkit uses a three-stage pipeline architecture:

1. **fetch_github_comments.py**: Uses GitHub CLI to fetch PR comments → outputs JSON

2. **parse_coderabbit_comments_v2.py**: Extracts CodeRabbit suggestions → outputs structured fixes JSON

3. **apply_coderabbit_fixes_v2.py**: Applies fixes to the codebase → modifies files

Fix Categories

The parser categorizes fixes into types:

Instructions

Running the Complete Pipeline

To process all CodeRabbit comments from a PR and apply fixes:

```bash

./coderabbit-fix OWNER/REPO PR_NUMBER

```

Or use the Python pipeline directly:

```bash

python coderabbit-tools/coderabbit_pipeline.py OWNER/REPO PR_NUMBER

```

Individual Tool Usage

Fetch comments from a PR:

```bash

./coderabbit-fetch OWNER/REPO PR_NUMBER

```

Parse CodeRabbit comments from saved JSON:

```bash

./coderabbit-parse comments.json

```

Apply fixes from parsed output:

```bash

./coderabbit-apply fixes.json

```

Dry-run mode (preview changes without applying):

```bash

python coderabbit-tools/apply_coderabbit_fixes_v2.py fixes.json --dry-run

```

Enhanced AI Formatter (Critical for Success)

The `coderabbit_ai_formatter.py` was enhanced to prevent Claude Code from:

**Key enhancements:**

Critical Success Patterns You MUST Follow

1. **Use TodoWrite tool** - Track every fix as a separate todo item. Manual tracking leads to incomplete work.

2. **Sequential processing** - Work on ONE issue at a time. NEVER work on multiple issues simultaneously.

3. **Validate after every change** - Run validation commands after EACH edit. Batching validation leads to hard-to-debug failures.

4. **File creation standards** - Always add EOF newlines. Ensure markdown files pass markdownlint.

5. **Verification checkpoints** - Get human approval before coding to prevent misinterpretation.

Validation Commands

ALWAYS run these commands after any changes:

```bash

make lint && coderabbit-linter && make validate && make test

```

The `coderabbit-linter` is a pre-commit linting wrapper for `coderabbit_linter.py`.

Adding New Fix Types

When extending the parser to handle new fix types:

1. Add the new category to `parse_coderabbit_comments_v2.py`

2. Update the fix detection logic in the same file

3. Ensure `apply_coderabbit_fixes_v2.py` can handle the new type

Testing Changes

Since there's no formal test suite, test changes by:

1. Using dry-run mode to preview changes

2. Testing on a sample PR with known CodeRabbit comments

3. Verifying JSON output at each pipeline stage

Shell Script Conventions

Optimized User Prompts

Two systematic prompts leverage the enhanced formatter:

**CodeRabbitPrompt.md**: For fixing existing CodeRabbit issues

**Github_Development_Prompt.md**: For implementing new GitHub issues

Requirements

Example Usage

```bash

Complete workflow for PR #123 in owner/repo

./coderabbit-fix owner/repo 123

Dry-run to preview changes

python coderabbit-tools/apply_coderabbit_fixes_v2.py fixes.json --dry-run

Validate all changes

make lint && coderabbit-linter && make validate && make test

```

Important Notes