This GitHub Action automatically reviews and applies accessibility improvements to your SwiftUI views using OpenAI's language models. It then creates a pull request with the suggested changes.

• Automated Accessibility Improvements: Leverages LLMs to identify and suggest accessibility enhancements for your SwiftUI code.
• Customizable OpenAI Model: Specify the OpenAI model you want to use (e.g., gpt-4, gpt-4.1, gpt-4-turbo). Defaults to gpt-4.1.
• Process Specific Files:
  • Process all SwiftUI files in the repository.
  • Process only SwiftUI files that were changed in the current commit.
• Pull Request Creation: Automatically creates a pull request with the applied changes.

This project was inspired by the talk and work of Allison McEntire on accessibility. You can find more about her insights here: gist.github.com/allisonpaigemcentire.

• Code Modification: This bot directly modifies your codebase by applying accessibility improvements. It is highly recommended that you carefully review all changes proposed in the pull requests created by this bot before merging them.
• Model Behavior: The quality and nature of the suggestions depend on the OpenAI model used. While the bot aims to apply only additive accessibility modifiers, always verify the changes.

1. OpenAI API Key: You need an OpenAI API key with access to the desired models. This key should be stored as a secret in your GitHub repository (e.g., OPENAI_API_KEY).
2. GitHub Token: The action uses a GitHub token to create branches and pull requests. The default GITHUB_TOKEN provided by GitHub Actions usually has the necessary permissions.

For this action to function correctly, it needs permissions to:

• Read repository contents.
• Write repository contents (to commit changes and push a new branch).
• Create pull requests.

You may need to adjust your repository's settings to grant these permissions to GitHub Actions. Navigate to your repository's Settings > Actions > General and ensure that "Workflow permissions" are set to "Read and write permissions". If you restrict the default GITHUB_TOKEN permissions, you might need to provide a token with sufficient scopes.

To use this action in your workflow, add a step similar to the following:

name: SwiftUI Accessibility Check

on: [push] # Or your preferred trigger

jobs:
  accessibility-bot:
    runs-on: ubuntu-latest
    permissions:
      contents: write      # Required to push new branches
      pull-requests: write # Required to create PRs
    steps:
      - uses: actions/checkout@v4
        with:
          # fetch-depth: 0 is recommended to fetch all history.
          # This is crucial if using `process_changed_files_only: true` 
          # or if your action needs to reliably determine the base branch for PRs.
          fetch-depth: 0 

      - name: Run SwiftUI Accessibility Bot
        uses: your-username/swiftui-accessibility-bot@v1 # Replace with your action's path or version
        env:
          GH_TOKEN: ${{ github.token }} # Pass the GitHub token for PR creation and push
        with:
          openai_api_key: ${{ secrets.OPENAI_API_KEY }}
          # Optional: Specify a different OpenAI model
          # openai_model_name: 'gpt-4-turbo'
          # Optional: Process only files changed in the current commit
          # process_changed_files_only: true

Refer to the action.yml file for a complete list of inputs and their descriptions.

• openai_api_key (required): Your OpenAI API key.
• openai_model_name (optional): The OpenAI model to use. Defaults to gpt-4.1.
• process_changed_files_only (optional): Set to true to only process files changed in the current commit. Defaults to false.

• branch-name: The name of the branch created with the accessibility improvements.
• files-modified: The number of files modified by the bot.

Contributions are welcome! Please open an issue or submit a pull request.

This project is licensed under the MIT License.