Review GitHub pull requests and label them using OpenAI GPT-4o-mini and Slack

Automate GitHub pull request reviews and labeling using OpenAI

This workflow automates the first line of code review for your development team. By leveraging OpenAI, it analyzes pull request diffs, assigns descriptive labels based on change size and category, posts summary comments back to GitHub, and keeps your team informed via Slack.

Who’s it for?

- DevOps Engineers looking to standardize PR triage. - Team Leads who want to provide instant feedback to developers. - Open Source Maintainers managing high volumes of contributions. - Development Teams aiming to reduce manual overhead in code reviews.

How it works / What it does

1. Trigger: The workflow starts via a GitHub PR Webhook when a pull request is opened or synchronized. 2. Data Gathering: It extracts PR metadata and uses the GitHub Node and HTTP Request Node to fetch a list of changed files and the raw code diff. 3. Analysis: A Code Node categorizes the changes (e.g., size labels like size/S or size/L). 4. AI Review: The AI Agent (powered by OpenAI) analyzes the code diff to generate a quality score, summary, and specific strengths/concerns. 5. Action: - The GitHub Node updates the PR with relevant labels. - An automated review comment is posted to the PR discussion. - A summary is sent to a Slack channel. 6. Reporting: All review data is logged into Google Sheets for long-term tracking and analytics.

Requirements

- GitHub Account: OAuth credentials with repository access. - OpenAI API Key: For the Chat Model (recommends GPT-4o-mini or higher). - Slack Workspace: A bot token to post to the code-reviews channel. - Google Sheets: A spreadsheet with headers matching the PR metadata.