GitHub Actions to Slack: Notify Only on CI Failures

You know, when you merge some code on main, a workflow fails...

...and nobody sees it until a customer or your teammate wonders why the migration did not run.

You can build a simple GitHub Actions Slack notification setup quickly using YAML and clicking in a few places.

Step 1: create a Slack incoming webhook

In Slack, create an incoming webhook for the channel where you want CI alerts.

• Go to your Slack app settings and enable Incoming Webhooks.
• Create a webhook and copy the URL.
• Keep it private. Treat it like a credential.

Step 2: save the webhook in GitHub secrets

In your repository, go to Settings -> Secrets and variables -> Actions, then add a new secret named SLACK_WEBHOOK_URL.

Do not hardcode this URL in your workflow file.

Step 3: add this GitHub Action

This uses slackapi/slack-github-action and posts a message only when the job fails.

name: CI with Slack Notification

on:
  push:
    branches: [main]
  pull_request:

jobs:
  test:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Run tests
        run: |
          echo "Run your test command here"
          # pnpm test

      - name: Notify Slack on failure
        if: failure()
        uses: slackapi/[email protected]
        with:
          payload: |
            {
              "text": "*${{ github.repository }}* - `${{ github.workflow }}` #${{ github.run_number }}\nStatus: *${{ job.status }}*\nBranch: `${{ github.ref_name }}`\nCommit: <${{ github.server_url }}/${{ github.repository }}/commit/${{ github.sha }}|${{ github.sha }}>"
            }
        env:
          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}
          SLACK_WEBHOOK_TYPE: INCOMING_WEBHOOK

A few practical notes

• Use if: failure() when you only want Slack alerts for failed jobs.
• If you want to run the step on both success and failure (but not on canceled jobs), use if: success() || failure().
• Avoid always() here because it also runs when the job is manually canceled.
• Keep messages short: status, branch, and a run or commit link.
• If the channel gets noisy, route by workflow (deploy alerts in one channel, test failures in another).
• For richer formatting (blocks, threads, user mentions), keep using the same action and evolve the payload.

Manual approach limits

• You have to repeat it for each repo, workflow, and job.
• It is manual to set up and maintain.
• You might forget one repository and miss critical alerts.
• You need secrets configured everywhere.
• Teams often need different webhook URLs and channels. In one repo, infra pipeline alerts may belong to one Slack channel while product pipeline alerts belong to another.