Frequently asked questions
The fine print, in plain words: everything here is already reflected in the price above.
How long until my project is fully up to date?
Almost never within the first month: the fee buys a batch of work every month, month after month. The more custom code and configuration your site carries, the more findings will turn up and the more months it takes to get everything up to date. A big project does not pay more each month: it pays more months. And when you want to shorten that road, that is exactly what the acceleration sprints are for.
What does my team have to do to get this running?
You hand me access to the Git repository and an up-to-date copy of the database, and that is it: I do the work and push it to a branch, and your team reviews it and merges whenever it suits them. You delegate it and stop thinking about it.
What is DruScan's continuous audit based on?
DruScan builds on Audit, a module I wrote and contributed to the Drupal community, free and open source: it audits the project and scores it. DruScan is the layer that turns those scores into history, alerts and a screen you can read without being a developer.
What exactly does the audit check?
It runs 24 audits and scores each area from 0 to 100: pending updates and security advisories, permissions and exposed accounts, caching at every level, database size, errors in the logs, fields and views set up wrong, technical SEO, images, multilingual and code quality with PHPStan, PHPCS and the automated tests. One detail that matters: the analysis runs on your server, and the only things sent to DruScan are the scores and the list of module versions. Your code, your content and your users never leave home. And there is no monthly report to wait for either: you see the same live panel I see.
Is there a lock-in or a minimum commitment?
No. It is a monthly fee you can walk away from whenever you want: the only ask is one month of notice, not a few days. I can afford that because the set-up is time I put in without billing it, and I only earn it back if you stay. So the pressure is on me, every single month: the service has to keep being worth more than it costs, or you leave.
How does billing work?
Billing is monthly and always in advance, before any work starts: I only work on months already paid. I am a freelance working alone, and charging in advance is what lets me spend my time on the projects instead of chasing payment for work already delivered. The set-up on my infrastructure comes included, at no separate cost. One-off extras, such as the sprints, go on their own invoice, separate from the fee, and can be booked whenever you need them. All prices are quoted without VAT. And everything is per project: an agency with several projects takes one plan for each.
Does the fee cover jumping to the next major Drupal version?
The fee covers the minor versions of the branch you are on: every 10.x if you run Drupal 10, every 11.x if you run Drupal 11. Jumping from 10 to 11, or from 11 to 12, is a different job, a major-version migration: the upgrade needs contrib compatibility checks, deprecated code replaced and many more tests, so it is quoted separately, at a closed price agreed before anything starts.
Do the tests guarantee an update will not break anything?
On every round I run the tests the project already has, and the more there are, the easier the two of us rest. But let me say it plainly: most projects arrive with very little coverage, and the tests cover the code written for you, not the community modules themselves. Updating twenty contrib modules at once is something no amount of tests can certify as risk-free. Which is why the branch reaches you for review, why the test coverage the plan includes is worth what it costs, and why on very large portals, where stability weighs more than running the latest of everything, it makes perfect sense to take only the security updates and, every few months, run a full round to bring the rest up to date.
Does anything reach production automatically?
Not from the update pipeline, not from any other tool in the process. The branch arrives reviewed and with the project tests run, but a human always presses the button.
Who runs the tests, and where?
Every test, PHPUnit and Playwright alike, I write and run on my own machine before delivering, so nothing broken reaches you. They all live in your repository, and anybody on your team can run them locally, on any branch. Continuous integration and scheduled runs on your own server are your team's to set up: on infrastructure of mine, nothing runs automatically.
Why does working without external AI models cost extra?
Part of my work leans on AI agents. When a project demands that not one line of code reaches an external model, for a strict NDA or a public-sector client, there is an option that keeps the analysis on models hosted on my own infrastructure, and it carries an extra monthly cost. The difference is not the hardware, it is the models. The powerful external ones, such as those from Anthropic or OpenAI, require sending data their way, but in exchange they are very capable: they leave the code in an acceptable state within few iterations before my review. The ones I can host locally are more limited and make more mistakes, and that with an infrastructure that is not small: we are talking several thousand euros invested. With them my human review needs more passes and more time on every task to guarantee the same quality. Running them costs me less than paying the external APIs, but that saving does not cover my extra hours, and that time is what shows up in the price.
Will one sprint leave my project fully fixed?
In each sprint I get through as much as two weeks allow, always starting with what matters most. How far that reaches depends on how much custom code the project carries and on the state it is in: on a small site it can mean clearing the whole list, and a big one takes several sprints. Whatever is left keeps moving through the monthly plan, and you can always book another one later.
