Automate PR Descriptions with AI Agents Using GitHub Actions

Writing pull request descriptions can be tedious, especially when you're focused on code. In this tutorial, you'll set up a GitHub Action that uses an AI agent (powered by OpenAI's GPT-4) to automatically generate a PR description from the diff. This workflow runs whenever a PR is opened or updated, saving you time and ensuring consistency.

Prerequisites

• Basic knowledge of GitHub Actions and YAML
• Python 3.9+ installed locally (for testing)
• An OpenAI API key (set as a repository secret)
• A GitHub repository with write access

Step 1: Create the Python Script

Create a file .github/scripts/generate_pr_description.py with the following code. It fetches the diff from the PR event, sends it to the AI, and writes the generated description to a file that the action will use.

import os
import json
import requests

# Get environment variables
diff = os.environ.get("DIFF", "")
api_key = os.environ.get("OPENAI_API_KEY")

if not diff or not api_key:
    print("Missing DIFF or OPENAI_API_KEY")
    exit(1)

prompt = f"""You are an expert code reviewer. Generate a concise pull request description from the following diff.
The description should include:
- Summary of changes
- Key modified files and their purpose
- Any potential impacts or considerations

Diff:
{diff[:6000]}
"""  # Truncate to avoid token limits

headers = {
    "Authorization": f"Bearer {api_key}",
    "Content-Type": "application/json"
}

data = {
    "model": "gpt-4",
    "messages": [
        {"role": "user", "content": prompt}
    ],
    "max_tokens": 500,
    "temperature": 0.3
}

response = requests.post(
    "https://api.openai.com/v1/chat/completions",
    headers=headers,
    json=data
)
response.raise_for_status()

description = response.json()["choices"][0]["message"]["content"]

# Save to file for later use
with open("/tmp/pr_description.txt", "w") as f:
    f.write(description)

print("Description generated successfully")

Step 2: Create the GitHub Action Workflow

Create .github/workflows/auto-pr-description.yml:

name: Auto PR Description

on:
  pull_request:
    types: [opened, synchronize]

jobs:
  generate-description:
    runs-on: ubuntu-latest
    permissions:
      contents: read
      pull-requests: write

    steps:
      - name: Checkout repository
        uses: actions/checkout@v4
        with:
          fetch-depth: 0  # Needed to get full diff

      - name: Get diff
        id: diff
        run: |
          git diff origin/${{ github.base_ref }}...${{ github.head_ref }} > /tmp/diff.txt
          cat /tmp/diff.txt
        shell: bash

      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: '3.10'

      - name: Install dependencies
        run: pip install requests

      - name: Generate PR description
        id: ai
        env:
          DIFF: ${{ steps.diff.outputs.diff_content }}
          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
        run: python .github/scripts/generate_pr_description.py

      - name: Update PR description
        env:
          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: |
          DESCRIPTION=$(cat /tmp/pr_description.txt)
          gh pr edit ${{ github.event.number }} --body "$DESCRIPTION"

Step 3: Fix the Diff Capture

The get diff step above will not save the diff to an output variable as written. Use this corrected version instead:

      - name: Get diff
        id: diff
        run: |
          DIFF_CONTENT=$(git diff origin/${{ github.base_ref }}...${{ github.head_ref }})
          echo "diff_content<> $GITHUB_OUTPUT
          echo "$DIFF_CONTENT" >> $GITHUB_OUTPUT
          echo "EOF" >> $GITHUB_OUTPUT
        shell: bash

Then update the Generate PR description step to use ${{ steps.diff.outputs.diff_content }} as the DIFF environment variable.

Step 4: Add the OpenAI API Key as a Secret

Go to your repository → Settings → Secrets and variables → Actions → New repository secret. Name it OPENAI_API_KEY and paste your key.

Step 5: Test the Workflow

Create a new branch, make a meaningful change, and open a PR. After a few seconds, you should see the PR description updated automatically with the AI-generated content.

Next Steps & Improvements

• Add error handling: if the API call fails, fall back to a default message.
• Use a cheaper model (e.g., gpt-3.5-turbo) for smaller diffs.
• Include structured labels: append “AI-generated: review required”.
• Integrate with your ticket system by parsing branch names (e.g., JIRA-123-feature).

This workflow saves at least 5 minutes per PR and ensures consistent descriptions across your team. Happy coding! 🚀