Taking over an existing web app: what I check first
The developer has moved on and the app still has to run. Here's the order I look at things in, before changing a single line.
A lot of my work starts the same way: there's an app that people rely on, the developer who built it has moved on, and something needs to change. Before I change anything, I want to know how it's put together and how it gets to production. Here's the order I look at things in.
1. Who holds the keys
Domain registrar, DNS, hosting, the database, the email service, payment accounts, the code repository. I make a list of every account the app depends on and check that the client, not a former developer, owns each one. It's the most important hour of the handover, and the one most often skipped. While I'm there I check that nothing has more access than it needs; I've written about least privilege before.
2. Can I run it?
Can I get the app running on my machine from what's in the repository? If the answer involves a file someone emailed once, that goes on the list to fix. An app that only one person can run is a risk to the business, not just to the next developer.
3. How does it get to production?
Is there a pipeline, or does someone copy files to a server? Is there a staging copy? Can a bad release be rolled back, and how long would that take? If deploying is scary, every change is slow, so making deploys boring is usually the first real improvement I make.
4. Backups, and whether they restore
A backup you've never restored is a hope, not a backup. I check what's backed up, where it goes, how often, and then I restore one somewhere safe to prove it works.
5. What's out of date
Framework and library versions, the language runtime, the operating system, TLS certificates. Some of it is urgent (security fixes), and some just needs a plan. I sort it into "today", "this week" and "this month".
6. Where it hurts
Error logs, slow pages, and the things users and staff complain about, like dead links or confusing browser error messages or dead end UI. That's where to spend effort first, not on the code I'd personally have written differently.
7. Rewrite or not?
Almost always: not yet. A working app holds years of decisions and edge cases that a rewrite will rediscover the hard way. In the past I've often had that moment when I realized that something that I did that I thought was making things simpler was in fact ignoring complexity that was actually inbuilt , and not accidental. The classic "Oh, that's why they did that" moment.
I'd rather make it safe to change, then improve it piece by piece. Especially now that analyzing the codebase has become easier with AI, and we can more speedily get to understand the wisdom inside a codebase that seems strange and probably could have been done better and more maintainably, but does contain an understanding of the company's business logic.
Sometimes a rewrite is the right call, as it was when I rebuilt the old PHP Computravel site in Nuxt, which was more about many new features than improving the old site. But that should be a decision made with the facts above in hand.
If you have an app that needs a new developer, I take on maintenance work. Email me and tell me what it is and what it runs on.