CI/CD & deployment
Send GitHub Actions alerts to on-call with CallHeim.
workflow_run webhook for CI run conclusions. CallHeim maps the payload, attaches repeats to the incident that is already open, and pages whoever is on call for the service it belongs to.
GitHub Actions workflow runs become CallHeim alerts: failed and timed-out runs are stored at P1, and runs that are still in progress or that succeed are stored at P4 (informational). Repeats on the same workflow collapse onto one incident, and the service the integration is bound to pages whoever is on call for it. CallHeim includes a payload mapping for GitHub Actions’ webhook format. GitHub signs its webhook itself, in a header and format CallHeim already checks, so turning signing on is a setting rather than extra setup.
How the alert reaches on-call
- Step 1A GitHub Actions workflow run in your repository changes state: requested, in progress, or completed.
- Step 2GitHub's workflow_run webhook POSTs the run, its workflow and its repository to your CallHeim ingest URL over HTTPS.
- Step 3CallHeim reads the run's conclusion (success, failure, cancelled, timed_out, …) and maps it to a severity; a run that has not completed is stored as informational regardless of any later conclusion.
- Step 4A repeat of the same workflow in the same repository attaches to the incident that is already open for it — the per-run id and run number are not part of the identity.
- Step 5The integration's bound service selects an escalation policy, which pages the on-call tier.
Setting it up
- 01
Create the GitHub Actions integration in CallHeim to get an ingest URL.
- 02
In the repository (or organization), go to Settings > Webhooks > Add webhook.
Paste the ingest URL into Payload URL and set Content type to application/json.
- 03
Under "Which events would you like to trigger this webhook?", select "Workflow runs" individually.
The default "Just the push event" sends nothing CallHeim can read here — you need the workflow_run event specifically.
- 04
Set a Secret on the GitHub webhook and turn signing on for the CallHeim integration, if you want requests verified.
GitHub signs every delivery with HMAC-SHA256 over the raw body using this secret, sent as X-Hub-Signature-256: sha256=<hex digest> — the exact header and format CallHeim checks (see Security below).
{
"action": "completed",
"workflow_run": {
"id": 5551212,
"name": "CI",
"head_branch": "main",
"status": "completed",
"conclusion": "failure",
"html_url": "https://github.com/example-org/web-store/actions/runs/5551212",
"run_number": 842
},
"repository": { "full_name": "example-org/web-store" },
"workflow": { "name": "CI" }
}What CallHeim reads from the payload
| Payload field | Maps to | Note |
|---|---|---|
| repository.full_name + workflow.name (or workflow_run.name) + head_branch + conclusion | incident title | Built from several fields joined together, not read from one. |
| workflow_run.conclusion | severity | failure or timed_out→P1, action_required→P2, cancelled or stale→P3, success/skipped/neutral→P4; when action is not completed (requested or in_progress), severity is always P4 regardless of any conclusion value present. |
| workflow_run.html_url, else run_number and status joined | incident body | |
| repository.full_name + workflow name | identity signal | The run id and run number are deliberately excluded, so consecutive failing runs of the same workflow collapse instead of opening a new incident every run. |
How CallHeim processes it
Routing
Each alert source is bound to a service, and the service’s escalation policy (or its team’s) sets who is paged.
Deduplication
A repeat of the same alert attaches to the incident that is still open for it in the same service, triggered or acknowledged, with no new incident and no new page. When the sender supplies its own dedup key (a PagerDuty dedup_key, an Opsgenie alias, a Datadog alert id, an Alertmanager fingerprint, or a dedupKey that is more than the alert title), this holds however late the repeat arrives. When the identity comes from the e-mail subject or the alert title, the repeat attaches only if the incident was triggered in the last 24 hours; after that it opens a new incident and pages. A repeat at P1 or P2 is not folded into an open P3, P4 or P5 incident: it opens a new incident and pages. Once an incident is resolved or closed, the next repeat opens a new incident. Identity is the repository and workflow name together — not the run id — so consecutive failures of the same workflow collapse onto one incident even though every run has its own id and run number. The repository and workflow name are the sender's own identifiers, not the alert title, so a repeat attaches at any age. While the incident is open, a repeat that attaches to it does not page again.
Recovery
GitHub Actions has no recovery event: a workflow_run notification describes one completed run, not a state that later clears. A later successful run is stored at P4 on the same workflow, and your team resolves the incident once a later run is known to have passed.
Troubleshooting
| Problem | Fix |
|---|---|
| No alerts arrive at all. | The webhook is likely still set to the default push event — open the webhook's settings and select "Workflow runs" explicitly under "Let me select individual events." |
| A run that's still in progress pages. | Expected — CallHeim reads every workflow_run delivery, including requested and in_progress; those are stored at P4 (informational) rather than skipped, so a running workflow does show up, just at low severity. Use an Orchestration Rule to route or re-prioritize them if you do not want them to open incidents. |
| A cancelled run pages at an unexpectedly high severity, or the reverse. | Confirm the delivery's action is completed and check the actual conclusion value delivered — cancelled or stale map to P3, not P1; only failure and timed_out map to P1. |
| HTTP 401 once you enabled signing. | Confirm the Secret set on the GitHub webhook is exactly the signing secret CallHeim generated for this integration — GitHub recomputes the signature per delivery, so a mismatched secret fails every request, not just the first. |
| Repeats from what look like unrelated workflows collapse together. | Shouldn't happen — identity includes the workflow name. Check workflow.name isn't empty in the delivered payload (it falls back to workflow_run.name, then to the literal word "workflow", which two differently-named-but-misconfigured workflows could both hit). |
Security
CallHeim verifies an HMAC-SHA256 signature on each source’s requests once you enable signing on that integration (the sending tool must be able to sign). GitHub signs each delivery itself with HMAC-SHA256 over the raw request body, sent as X-Hub-Signature-256: sha256=<hex digest>. CallHeim accepts that header and format as well as X-CallHeim-Signature: paste the signing secret CallHeim generates for this integration into the webhook's Secret field on GitHub's side and turn signing on. Until you do, the integration URL is the credential: keep it secret, and if it is exposed, disable the integration and create a new one.
Vendor documentation checked
- GitHub: workflow_run webhook event — checked 2026-09-24
- GitHub: Validating webhook deliveries — checked 2026-09-24
The thresholds it passes through
- Repeat, sender-supplied dedup key
- attaches, any age
- Repeat, title or subject only
- attaches, incident under 24 h old
- Flap threshold
- 4 transitions / 600s
- Title correlation
- similarity ≥ 0.6, same source and service
- Group window default
- 600s
All defaults are published. You can turn title correlation off or change its threshold, and set the window on your own noise rules; the dedup rule and the flap settings are fixed. A P1 or P2 repeat on an open P3 to P5 incident opens a new incident and pages. How alerts are processed →
Point GitHub Actions at CallHeim and see what it does with your alerts.
Start on the free trial, connect GitHub Actions and send a test alert to see the incident it opens.