mybug.dev

Night-shift on-call for startups and SaaS without an SRE

We watch your production while your team sleeps.

When an alert fires at 3 a.m., we pick it up. We triage it, fix it if it is safe and approved, or wake the right person with the facts already written down. Every morning you get a short report of what happened overnight.

  • 15 minnight-shift response
  • Asyncno calls required
  • Read-onlyaccess by default
incident · api-gateway · 5xx rate resolved
SeveritySEV-2
Time to ack3m
Team pagedNo
  1. Alert fired

    5xx rate above 4% on /checkout for 5 min.

  2. Acknowledged

    First assessment posted in your channel.

  3. Mitigated

    Runbook step 3: roll back last deploy. Approved in advance.

  4. Morning report

    Timeline, cause, and follow-ups for your team.

Illustrative example

The problem

Production does not sleep. Your team has to.

Small teams without an SRE usually end up with the same three nights on repeat.

Alerts nobody sees

The pager fires at 3 a.m. in your timezone. Customers notice before anyone on the team does.

The founder is the on-call

One or two people carry every night. Sleep gets worse, and so does the daytime work.

Noise hides the real thing

Flaky alerts train people to ignore the pager, until the one alert that mattered.

Services

A night shift, with a clear scope.

Scope is capped on purpose, so it stays predictable for both sides.

Included

  • Night-shift alert monitoring for your timezone
  • Triage of every alert: real, noise, or needs a human
  • Fixes from your runbook, with your approval
  • Escalation to your on-call person, with context
  • A written morning report
  • Notes on noisy alerts worth tuning or removing

Not included

  • Feature work or code changes in your product
  • Infrastructure migrations and rebuilds
  • Daytime support (your team owns the day)
  • Unapproved changes to production
  • Compliance certification work

Anything outside the scope is quoted hourly up front, and only done if you say yes.

How it works

Async from day one. Written, not spoken.

No calls needed. Email, Slack, and written reports.

  1. 01

    Email us

    Tell us your stack, your timezone, and what currently wakes people up.

  2. 02

    NDA first

    We sign your NDA (or send ours) before we ask for any access.

  3. 03

    Read-only access

    Dashboards, logs, alerting. Nothing that can change production unless you decide otherwise.

  4. 04

    Runbook, in writing

    What counts as urgent, who to escalate to, what we may fix on our own.

  5. 05

    We take the night shift

    Alerts route to us. You get a written report each morning.

15min

15-minute response during our night shift for your timezone.

If we miss it, that month is free. Response means a human has looked at the alert and posted a first assessment in your channel.

Async-first

No calls needed. Email, Slack, and written reports.

Read-only by default

Changes only with your explicit approval.

NDA before access

Signed before we see a single log line.

Cancel anytime

No lock-in. The first month is free.

Sample audit

See how we write before you give us anything.

Before a trial we can run a free external check: only what anyone on the internet can see. No access needed. Here is an anonymized example of what comes back.

Example: B2B SaaS, EU · external check, no access granted
SeverityFindingSuggested fix
HighAPI origin host reachable directly, bypassing the CDN, and returns server version banners.Allow origin traffic only from the CDN; hide version headers.
MediumDMARC set to monitor-only (p=none) with no reporting address, so spoofing goes unseen.Add a report mailbox, check alignment, then move to quarantine.
MediumMissing security headers: no Content-Security-Policy, no frame protection.Add CSP and frame-ancestors / X-Frame-Options.
MediumSLA advertised on the pricing page, but no public status page.Publish a status page fed by real uptime checks.
OKTLS configuration is sound: modern protocols and a valid certificate.Nothing to do.

Pricing

Start free. Decide after a month.

The trial is the same service as a paid month. No card needed to start.

Trial

Free for 1 month

  • Same service as a paid month
  • 15-minute response promise applies
  • No card needed to start
  • Not continuing? We hand back access and delete what we hold
Start a 1-month free trial

After the trial

Written quote

  • Quoted for your stack and your night-shift hours
  • Cancel anytime, no lock-in
  • If we miss the 15-minute response, that month is free
  • Out-of-scope work quoted hourly up front, only if you say yes
Ask for a quote

FAQ

Questions, answered in writing.

Anything else: hi@mybug.dev. We reply in writing.

What access do you need, and how is it kept safe?

Read-only access to monitoring, logs, and alerting, granted after an NDA is signed. Use separate accounts or scoped tokens so you can see and revoke our access at any time. We do not ask for production write credentials by default.

What if you break something?

With read-only access, we can't. Any change happens only with your approval, following steps written down in advance in the runbook. If something is unclear at night, we escalate rather than guess.

Which timezone do you cover?

Yours. We agree the night-shift hours for your timezone in writing before the trial starts. The 15-minute response promise applies inside those hours.

How does the free trial work?

One month, free, same service as a paid month. No card needed to start. At the end you decide to continue or not; if not, we hand back access and delete what we hold.

Do we need meetings?

No. Everything works over email, Slack, and written reports. A call is fine if you want one, but it is never required.

Sleep through the night.

Email us with your stack and timezone. We reply in writing.