Receive GitHub Actions build failures and deployment results on your phone
Receive CI and deployment notifications
Send GitHub Actions results to iPhone, Android, and the web when a workflow finishes or a build needs attention. A curl step calls the Signalgrid Push API directly, so you can use your existing workflows without installing a third-party notification action.
Prerequisites
- A GitHub repository where you can manage Actions workflows and repository secrets.
- A GitHub-hosted Ubuntu runner, or a runner with Bash, curl and jq installed.
- A Signalgrid account with an iPhone, Android device, or web client registered to receive notifications.
- Your Signalgrid client key and a channel token. Use the channel token rather than the channel name.
- Outbound HTTPS access to
https://api.signalgrid.co.
Add the repository secrets
In the repository, open Settings → Secrets and variables → Actions and add two repository secrets:
SIGNALGRID_CLIENT_KEYSIGNALGRID_CHANNELKeep credentials in secrets rather than committing them to the workflow. Use environment variables for notification text, as in the examples, instead of inserting event text directly into shell commands.
Create a manual notification workflow
Save this file as .github/workflows/signalgrid-test.yml on your default branch:
Open the repository's Actions tab, select Signalgrid notification test and choose Run workflow. Check the run result and the notification on your device. The script checks both HTTP errors and Signalgrid's application response code.
Notify when a workflow step fails
Add the following step at the end of the existing job you want to monitor. Keep it inside that job's steps: list and match the surrounding indentation. The failure condition runs it after an earlier step in the job fails.
The message includes the workflow run URL so you can open the failure details. For a completion notification, use the same step with if: ${{ success() }}, set the type to SUCCESS and adjust the title. Place it after the build or deployment steps whose success you want to report.
Jobs and forked pull requests
A step reports failures in its own job. Add it to each job you want to monitor, or use a separate notification job with explicit dependencies. Repository secrets are normally unavailable to workflows triggered by pull requests from forks; use trusted push or manual runs for the initial test.
Send critical alerts
For urgent incidents, set critical to true and type to CRIT. These are separate settings: the visual notification type alone does not enable critical delivery. Critical notifications require a Single or Teams subscription and the appropriate device permissions. See Critical Notifications for iOS Critical Alerts and Android notification settings. Keep routine results non-critical.
In the request, change type=WARN to type=CRIT and critical=false to critical=true for a failure that needs immediate attention.
Troubleshooting
- Run workflow is missing: check that the manual workflow is on the default branch and declares
workflow_dispatch. - The step fails: verify both repository secret names, the API response and runner access to Signalgrid.
- No failure notification: check the step is in the job that failed and uses
failure(). Errors marked continue-on-error may not fail the job. - Notification step is skipped on a fork: check the event and secret availability. Do not expose credentials to untrusted workflow code.
- Repeated messages: matrix jobs and workflow reruns can each send a notification; place or condition the step accordingly.