Nightlamp Health-Check Recipes

Nightlamp Health-Check Recipes

Copy-paste monitoring recipes for the failure modes that quietly break no-code and AI-built apps — checkout, webhooks, auth, DNS/SSL, cron, lead forms, third-party APIs, and visual regressions.

The problem: generic uptime monitoring tells you the homepage returns 200. It doesn't tell you the checkout button stopped working, the Stripe live webhook silently went nowhere, or the cert on a forgotten subdomain expires next week. This repo is built for no-code founders and small teams: each recipe catches one money-losing failure mode and gives your on-call a triage checklist, so you can fix it before customers notice.

Each recipe is a self-contained folder: a plain-English runbook plus a machine-readable nightlamp.recipe.yaml you can copy into Nightlamp, Checkly, Grafana, Prometheus blackbox_exporter, or any tool that speaks the same check primitives. No install, no runtime, no lock-in.

At a glance

Use this if…

• You shipped a flow on Bubble, Lovable, Cursor, Webflow, FlutterFlow, or similar, and want to know it broke before the customer tells you.
• You run cron jobs, webhooks, or third-party integrations that can fail silently.
• You want a triage checklist ready in the runbook, not invented at 2am.

Probably not for you if…

• You only need basic "is the homepage up?" uptime checks — any free uptime tool covers that.
• You have a full SRE/observability team with bespoke synthetic tests already.
• You need a hosted monitoring product out of the box — these are the recipes; Nightlamp is one way to run them continuously.

Contents

• Available recipes
• Quickstart
  • With Nightlamp
  • With any other tool
• What's in a recipe
• Repository layout
• Editing & validating recipes
• Contributing
• Governance and support

Available recipes

Recipes are organized by scenario (the thing that actually breaks), not by check primitive — so you start from the symptom your customer reports.

Quickstart

There's nothing to compile or install to use a recipe. The whole flow is: read the runbook → copy the YAML → swap in your values → create the monitors.

Use a recipe with Nightlamp

Every recipe in this repository has a one-click onboarding path in the Nightlamp recipe catalogue. Pick the flow, sign in, confirm the app URL, and Nightlamp creates the app plus the recipe's primary monitor — or opens the monitor form with the recipe's fields prefilled. The trial is 14 days with no card. The table above links each recipe straight to that path.

1. Open the recipe's nightlamp.recipe.yaml (e.g. recipes/ssl-expiry/nightlamp.recipe.yaml).
2. Replace {{...}} placeholders with your domain, paths, and non-secret identifiers.
3. Create one monitor per checks[*] entry in Nightlamp (or paste the monitor JSON — most check types map 1:1 to monitor fields).
4. Copy the Triage section from the recipe README.md into your runbook so on-call has a starting checklist.

✅ Expected result: within one check cadence you have live monitors that fail specifically on the failure mode — not just on a downed homepage — and a written triage path for whoever gets paged.

Use a recipe with any other tool

The manifest names monitor types using a generic taxonomy, so they map onto any synthetic-monitoring tool:

Map these onto Checkly browser/API/heartbeat checks, Grafana Synthetic Monitoring, Prometheus blackbox_exporter, Upptime, or whatever you already run.

Supported platforms: recipes are platform-neutral YAML + Markdown and work on macOS, Linux, and Windows.

What's in a recipe

Every recipe folder gives the on-call human a complete story:

• Symptom — what the customer actually reports.
• Failure modes — the likely root causes, ranked.
• Checks — the exact monitor(s) to set up, with concrete YAML.
• Expected signals — what pass and fail look like.
• False positives — what looks broken but isn't.
• Triage — a short checklist for whoever gets paged.
• Alert routing — severity, channel, and when to suppress.

Repository layout

recipes/
  <scenario>/
    README.md                 # symptom, failure modes, triage, alerting
    nightlamp.recipe.yaml     # machine-readable manifest (monitors + metadata)
schemas/
  recipe.schema.json          # JSON schema for the manifest
scripts/
  validate_recipes.py         # local validator (runs in CI too)
.github/
  ISSUE_TEMPLATE/             # recipe-request, recipe-bug, ...
  workflows/                  # recipe validation + OSS compliance

Editing & validating recipes

You only need this if you're editing a recipe — reading and copying one needs nothing installed.

git clone https://github.com/GipsyChef/nightlamp-health-recipes.git
cd nightlamp-health-recipes
python3 -m pip install pyyaml jsonschema
python3 scripts/validate_recipes.py

Prerequisites: Python 3.10+, pyyaml, jsonschema. The same validator runs in CI on every pull request.

Contributing

We welcome:

• New recipes for failure modes you've actually hit.
• Updates when a vendor behavior, URL, or workaround changes.
• Refinements to triage steps, alert routing, or false-positive notes.

See CONTRIBUTING.md for the recipe format, validation rules, and what we won't merge (secrets, customer-identifying details, unsafe probing patterns).

Governance and support

• License: Apache-2.0 — yours to use, fork, and adapt.
• Maintainers: MAINTAINERS.md
• Security disclosure: SECURITY.md
• Code of conduct: CODE_OF_CONDUCT.md
• Support boundaries: SUPPORT.md
• Recipes are reviewed quarterly; each has a last_reviewed date in its manifest.

This is a community knowledge base. The recipes are yours to run anywhere — Nightlamp is the easiest way to run, alert on, and learn from them continuously.