Skip to content
Sun Software logo with the tagline Engineering Beyond Code
Engineering Standards

The rules we work to on every release.

Want to know how your project will be built? These are the rules our engineers follow on every custom software and AI project. We publish them so you can check our work against them.

release-recordproduction
  • Architecture reviewSigned off
  • Second-engineer code reviewApproved
  • Automated testsPassing
  • Staging releaseVerified
  • Production monitoringAlerts armed
  • Handover documentationComplete

Why do we publish our rules?

Most companies say they care about quality. We'd rather show you the rules, the record each one leaves behind and where you can look for it. If we fall short of anything on this page during your project, tell us.

The seven standards

Seven rules, and what you can check for each.

  1. 01

    Every change is reviewed

    Each change to your code is sent as a pull request, which is a request to add it to the main version. The main version is locked, so nothing goes in until a second engineer has read and approved it. Code written with AI tools gets the same review as code written by hand. For you, it means a second person has checked each change before it goes in.

    What you can inspect

    The branch protection settings and the review history in your repository.

  2. 02

    Automated tests on every change

    We write automated tests for the parts of your product your business depends on most. They run every time the code changes, and a failing test stops that change from going in. The aim is to catch a change that breaks something that used to work before your customers find it. At kickoff we agree with you how much of the code the tests should cover, because the right number differs by project.

    What you can inspect

    The history of test runs in the build pipeline.

  3. 03

    Security checks before release

    Before release, we check every web app against a baseline drawn from OWASP ASVS Level 1, a public list of security requirements for web apps. We check AI features against the OWASP Top 10 for LLM Applications, a public list of the most common risks in AI features. Every change is scanned for passwords or keys left in the code and for third-party code with known weaknesses. Each person gets only the access their job needs, and every company account uses two-step login.

    What you can inspect

    The signed security checklist for each release.

  4. 04

    A test copy before going live

    Changes go first to a staging environment, a private copy set up like your live system. That way problems show up on the copy and not in front of your customers. Every release has a checklist and a plan for going back to the previous version if something goes wrong. Our technical lead signs off major releases.

    What you can inspect

    The release record.

  5. 05

    Someone watching after launch

    After launch, one named person is responsible for uptime, errors and alerts, so you know who acts when something breaks. For AI systems we also keep a log of what the AI produces and track what it costs to run. Whenever a model or prompt changes, we re-run a fixed set of test questions to see whether the answers got better or worse.

    What you can inspect

    The dashboards and the alert history.

  6. 06

    A handover you can run with

    Every project ends with a README (a guide to the code), a diagram of how the parts fit together, a runbook for running and fixing the system, and a record of the key decisions. We hand over passwords and keys through a password manager. You can run it yourself or take it to another team. Nothing depends on us staying.

    What you can inspect

    The handover checklist.

  7. 07

    Regular upkeep

    On systems we look after on a retainer, we update the third-party code every month and test restoring from a backup every quarter. A backup only helps if it restores, so we try it. We also write up every incident, so you can read what happened and what we changed.

    What you can inspect

    The maintenance log and the incident notes.

AI tools

How do we use AI tools?

We use AI tools to help write, review and test your code. We do it under a written policy that every employee and contractor signs.

  • We use approved tools only, on plans that do not train on your code.
  • No passwords, keys or personal data go into any AI tool.
  • An engineer answers for every change, and a second engineer reviews it.
  • AI agents never hold the passwords or keys to your live system, and never merge or deploy on their own.
  • App builders are for prototypes. Nothing goes live without a full review.
  • You can ask us not to use AI tools on your project.
  • If something goes wrong with your data, we tell you.

Ask us for the full policy and our current list of approved tools.

Quality reporting

What do we report to you?

If you're on a retainer, you get a quality report every month. It covers seven things.

  • Deployments made
  • Change failure rate
  • Time to restore service
  • Lead time for changes
  • Open vulnerabilities, by severity
  • Uptime
  • Incidents, with a write-up for each

The first four are the DORA metrics, a widely used set of measures for software delivery. In plain words: how often we release, how often a release causes a problem, how fast we fix it and how long a change takes to go live.

Your code and data

How do we protect your code and data?

  • We sign an NDA before we see anything.
  • We work in your repositories and accounts wherever possible, so they stay yours.
  • Each person has their own named account, and we keep a list of who can access what.
  • Your live data stays off our laptops. We develop and test with made-up data.
  • When the project ends we delete or return your data, and confirm it in writing if you ask.
  • A data processing agreement is available.
We do not yet hold SOC 2 or ISO 27001 certification. Our controls are written down, and you can review them before you sign.

What don't these standards promise?

No process removes every defect. These rules make problems rarer, easier to spot early and cheaper to fix. When something does go wrong, we tell you and we write up the cause.
Questions

Questions about our standards.

Do these apply to WordPress and Shopify work?

Partly. Every custom software and AI project follows all seven. WordPress and Shopify builds follow the same review, staging and handover rules, adapted to the platform.

Can we see evidence before we sign?

Yes. Ask for our release checklist, a pull-request review with the client details removed and a sample audit report.

Do your contractors follow the same rules?

Yes. Contractors sign the same policy and their work goes through the same review.

Can you work to our standards instead?

Yes. Where yours are stricter, yours apply.

What happens when a standard cannot be met?

The reviewer blocks the release. Any exception is written down, approved by our technical lead and shared with you.

Hold us to it.

Ask for our release checklist and a sample audit report, or book an audit of your own app.

  • The first conversation is free
  • NDA before you share anything
  • A reply within one business day