Deploy Verification Loop
FeaturedScheduledAbout Deploy Verification Loop
This loop verifies a deployment once it has gone out. Each round requests the health URL with curl and looks at whether every configured endpoint succeeds. If not, it waits for the next interval and tries again, up to 8 rounds.
Key features
- 01Interval trigger repeats the check without manual reruns
- 02Uses curl -fsS, so HTTP errors count as failures
- 03Covers post-deploy health and smoke endpoints
- 04Capped at 8 rounds before retrying stops
Use cases
- 01Confirming a service is healthy after a new release
- 02Waiting for instances to become ready during a rollout
- 03Running a smoke check on key endpoints after publishing
- Goal
- All post-deploy health and smoke endpoints return success.
- Exit when
- Stop when every configured endpoint succeeds.
- Between iterations run
- curl -fsS <your-health-url>
Launch prompt
Paste into Codex / Cursor. The agent repeats the steps below until the exit condition is met.
Start the "Deploy Verification Loop" loop. Goal: All post-deploy health and smoke endpoints return success. Max iterations: 8. Step 1: Request the health endpoint — Run curl -fsS https://your-app.example/health. A non-zero exit status or an HTTP error counts as a failed check for this round. `curl -fsS https://your-app.example/health` Step 2: Collect the results — Note whether each configured endpoint succeeded or failed, and keep the reason for any failure. Step 3: Decide whether to continue — If every configured endpoint succeeded, end the loop. Otherwise wait for the next interval and start another round, until the limit of 8 rounds is reached. Between iterations run: `curl -fsS <your-health-url>` Exit when: Stop when every configured endpoint succeeds.
Steps
Each round runs in order; if the exit condition isn’t met, another round starts, up to 8.
Run curl -fsS https://your-app.example/health. A non-zero exit status or an HTTP error counts as a failed check for this round.
curl -fsS https://your-app.example/healthNote whether each configured endpoint succeeded or failed, and keep the reason for any failure.
If every configured endpoint succeeded, end the loop. Otherwise wait for the next interval and start another round, until the limit of 8 rounds is reached.
Supported agents
Coding agents that can run this loop directly.