A GitHub Action that extracts Jira tickets from commits and sends Slack notifications for frontend staging deployments.

• Extracts Jira tickets (IL-XXXXX) from commit messages
• Creates or updates Slack messages per branch
• Threads subsequent deployments to existing messages
• Replaces staging URL with latest (no accumulation)
• Deduplicates Jira tickets across the entire thread
• Non-fatal error handling (won't fail your CI)
• Built-in retry logic for Slack API calls

- uses: ./actions/fe-staging-notification
  with:
    repo: "Mito"
    network: "mainnet"
    staging_url: "https://staging.example.com"
    slack-user-token: ${{ secrets.SLACK_USER_TOKEN }}
    slack-bot-token: ${{ secrets.SLACK_BOT_TOKEN }}

This action uses a dual-token approach:

The user token is required because the Slack search.messages API only works with user tokens, not bot tokens.

fe-staging-notification/
├── action.yml              # Action definition (node24)
├── action.bash.yml         # Backup of original bash version (for rollback)
├── package.json            # Dependencies and scripts
├── vitest.config.js        # Test configuration
│
├── src/                    # Source code
│   ├── index.js            # Main entry point - orchestrates the action
│   ├── git.js              # Branch name detection
│   ├── jira.js             # Jira ticket extraction from commits
│   └── slack.js            # Slack API helpers with retry logic
│
├── __tests__/              # Unit tests (Vitest)
│   ├── git.test.js
│   ├── jira.test.js
│   └── slack.test.js
│
├── dist/                   # Bundled output (auto-generated)
│   └── index.js            # Single-file bundle (~960KB)
│
├── .husky/                 # Git hooks
│   ├── _/husky.sh
│   └── pre-commit          # Runs tests + build before commit
│
├── coverage/               # Test coverage reports (gitignored)
├── node_modules/           # Dependencies (gitignored)
│
├── README.md               # This file
└── MIGRATION.md            # Migration guide from bash to JS

┌─────────────────────────────────────────────────────────────┐
│                        index.js                              │
│                    (Main Orchestrator)                       │
│                                                              │
│  1. Get inputs from action.yml                               │
│  2. Detect branch name                                       │
│  3. Extract Jira tickets from commits                        │
│  4. Search for existing Slack message                        │
│  5. Update existing OR create new message                    │
│  6. Set outputs                                              │
└─────────────────────────────────────────────────────────────┘
           │                │                │
           ▼                ▼                ▼
    ┌──────────┐     ┌──────────┐     ┌──────────┐
    │  git.js  │     │ jira.js  │     │ slack.js │
    └──────────┘     └──────────┘     └──────────┘
    
    getBranchName()   extractJiraTickets()   slackRequest()
    - INPUT_BRANCH    - git fetch origin/dev  - Retry logic (3x)
    - Event file      - git log               - Rate limiting
    - GITHUB_HEAD_REF - Pattern matching      
    - GITHUB_REF_NAME generateJiraLinks()    searchExistingMessage()
                      - Slack formatting      updateMessage()
                                              postMessage()
                                              postThreadReply()
                                              addMessageId()

1. Dual Slack Tokens: User token for search (API limitation), bot token for posting
2. 30-Day Search Limit: Prevents finding stale messages from old deployments
3. Thread-Based Deduplication: Scans entire thread for Jira tickets, not just main message
4. URL Replacement: Shows only the latest staging URL (previous fix for URL accumulation)
5. Non-Fatal Errors: Logs warnings but never fails the CI pipeline
6. Retry Logic: 3 retries with 2-second delays for transient Slack errors

• Node.js 20+
• npm

cd actions/fe-staging-notification
npm install

The .husky/pre-commit hook automatically:

1. Runs lint-staged (tests related files)
2. Runs npm run build to regenerate the bundle
3. Stages dist/index.js if it changed

This ensures:

• Tests pass before every commit
• The bundled output is always up-to-date
• No broken code gets pushed

1. Modify source files in src/
2. Add tests in __tests__/
3. Run npm test to verify
4. Commit (pre-commit hook will build automatically)

• Create JavaScript source files
• Write unit tests (45 tests)
• Configure Vitest with coverage thresholds
• Set up Husky pre-commit hooks
• Build and verify bundle
• Backup bash version to action.bash.yml

Run both versions in parallel to verify parity:

# In your workflow, temporarily run both:
- name: Notify (JS version)
  uses: ./actions/fe-staging-notification
  with:
    repo: ${{ inputs.repo }}
    # ... other inputs

- name: Notify (Bash version - shadow)
  uses: ./actions/fe-staging-notification/action.bash.yml
  with:
    repo: ${{ inputs.repo }}
    # ... other inputs

Compare outputs for 1-2 weeks before proceeding.

The repository is now configured to use the JS version:

No workflow changes required - the action interface is identical.

After deployment, monitor for:

1. Slack messages appearing correctly
2. Jira tickets being extracted
3. Thread replies working
4. No CI failures due to the action

Once confident:

# Remove the backup
rm action.bash.yml
git add -A
git commit -m "chore: remove bash backup after successful JS migration"

If issues arise in production, rollback takes less than 5 minutes:

cd actions/fe-staging-notification

# Restore bash version
cp action.bash.yml action.yml

# Commit and push
git add action.yml
git commit -m "revert: rollback fe-staging-notification to bash version"
git push

After pushing, trigger a test deployment and verify:

• Slack message appears
• Jira tickets are extracted
• Thread behavior works

If rollback was needed, investigate:

1. Check GitHub Actions logs for errors
2. Compare JS output vs expected bash output
3. Test locally with node dist/index.js (mock inputs)
4. Add regression tests before re-attempting migration

To see detailed logs, check the GitHub Actions output. The action logs:

• Branch name detected
• Number of Jira tickets found
• Whether existing message was found
• Success/failure of Slack API calls

# Set environment variables
export INPUT_REPO="TestRepo"
export INPUT_NETWORK="testnet"
export INPUT_STAGING_URL="https://test.example.com"
export INPUT_SLACK_USER_TOKEN="xoxp-..."
export INPUT_SLACK_BOT_TOKEN="xoxb-..."
export GITHUB_ACTOR="testuser"
export GITHUB_REF_NAME="feat/test-branch"

# Run the action
node dist/index.js

• Migrated from 542-line bash script to JavaScript
• Added retry logic for Slack API calls (3 retries, 2s delay)
• Added unit tests (45 tests, 50%+ coverage)
• Added pre-commit hooks for test validation
• Fixed: Staging URL now replaces instead of accumulates
• Fixed: Message ID added to new messages for reference
• No breaking changes to action interface

• Original bash implementation
• Preserved in action.bash.yml for rollback