Run GitHub Actions from Jira Automation

You can start a GitHub Actions workflow from a Jira work item with Jira Automation alone, no app needed. This page shows the setup, how to get the result back onto the work item, and what's still missing afterwards.

1. Let the workflow be started from outside

Add a workflow_dispatch trigger with the inputs you want to pass from Jira. Declare them as strings.

# .github/workflows/deploy.yml
on:
  workflow_dispatch:
    inputs:
      issue_key: { type: string, required: true }
      environment: { type: string, required: true }

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v7
      - name: Deploy
        env:
          ENVIRONMENT: ${{ inputs.environment }}
        run: ./scripts/deploy.sh "$ENVIRONMENT"

2. Create a GitHub token

Automation calls the GitHub API with a token. Use a fine-grained personal access token limited to the one repository, with Actions: Read and write. A classic token needs the repo scope, which is much broader.

3. Build the rule

  1. Trigger: Manual trigger from work item. Pick the groups that may run it under Groups that can run trigger. To ask for an environment, turn on Get input from users and add a field named environment.
  2. Action: Send web request.
    • URL: https://api.github.com/repos/OWNER/REPO/actions/workflows/deploy.yml/dispatches
    • HTTP method: POST
    • Headers: Accept: application/vnd.github+json and Authorization: Bearer <token>. Tick Hidden on the token so it shows as asterisks after saving.
    • Web request body: Custom data
{
  "ref": "main",
  "inputs": {
    "issue_key": "{{issue.key}}",
    "environment": "{{userInputs.environment}}"
  }
}

People now start the workflow from the work item's Actions menu.

4. Get the result back

Automation only starts the run. For the work item to hear how it went, the workflow has to call the Jira REST API itself, with a Jira API token saved as a GitHub secret. For example, as the last step:

- name: Comment on the work item
  if: always()
  env:
    JIRA_SITE: ${{ secrets.JIRA_SITE }}
    JIRA_AUTH: ${{ secrets.JIRA_EMAIL }}:${{ secrets.JIRA_API_TOKEN }}
    ISSUE_KEY: ${{ inputs.issue_key }}
    TEXT: 'Deploy ${{ job.status }}: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}'
  run: |
    jq -n --arg t "$TEXT" \
      '{body: {type: "doc", version: 1, content: [{type: "paragraph", content: [{type: "text", text: $t}]}]}}' |
    curl -sf -u "$JIRA_AUTH" -H 'Content-Type: application/json' \
      -X POST "https://$JIRA_SITE/rest/api/3/issue/$ISSUE_KEY/comment" --data @-

What this setup leaves out

  • The next step. After a staging deploy, nothing offers Promote to production to the release manager or Roll back to the developers. Someone has to know which rule to run next.
  • The run's state on the work item. It shows only the comments the workflow writes. There's no running, passed or failed state to look at.
  • Tokens in two places. A GitHub token in the rule and a Jira token in GitHub, each owned by someone.
  • Usage. Each run of the rule counts toward your Automation usage limit.

The same loop with Cuelight

Cuelight for Jira keeps your workflow and replaces steps 2 to 4. An admin connects GitHub once, and actions define the button, its form and who can press it. The workflow reports back with one step, and chooses which buttons come next:

- name: Report to Jira
  if: always()
  uses: huseyinbuyukdere/cuelight-report@v1
  with:
    status: ${{ job.status }}
    next: promote-production, rollback-staging

Make the action and the workflow YAML in the action builder, or see the whole loop in the demo.