My job feed broke silently for 3 days. So I built a CI guardrail that fails the build when it happens again

Description and use

The jobfeed-watchdog script automatically checks JSON feed structure in CI and fails the build on regression, providing instant alerts for silent failures.

Problem

A job feed returned an empty data field for three days without any alerts, causing dashboards to show zero jobs while the API was technically up.

Outcome

The CI guardrail automatically fails builds when a JSON feed’s structure regresses, providing immediate alerts and detailed diffs, preventing silent data loss.

Workflow

• Install jobfeed-watchdog via pip or add the GitHub Action to .github/workflows
• Define feeds.json with endpoint URLs
• Configure the Action with feeds.json and min_records threshold
• Schedule the Action to run periodically
• On each run, the Action fingerprints the feed structure and compares it to the last known‑good state
• If a regression is detected, the Action fails the build and logs the specific signal that broke

Evidence

My job feed broke silently for three days before anyone noticed. The board's API hadn't gone down — it had just started returning an empty data field where there used to be a list.

Source

jobfeed-watchdog runs a structural fingerprint on a scheduled basis against any JSON/YAML endpoint that returns a list of records... turns the Actions run red when something regresses.

Source

The whole thing is 26 tests, all passing, stdlib-only: $ python -m pytest .............. 26 passed in 0.36s

Source