Urgent takeover
When communication breaks down, the first priority is securing access to repositories, cloud services, and domains before engaging a replacement developer for a structured audit.
Immediately revoke former developer permissions and secure all master credentials (GitHub, AWS, Stripe, Vercel).
Have a verified architect evaluate whether the current repository code matches what was deployed.
Define an itemized task backlog and resume development with transparent daily progress.
Silence from a developer is uncomfortable but not usually catastrophic. Losing access to the accounts your product runs on is catastrophic, and it happens quietly while you are still sending polite follow-up emails.
Work through the list below today, regardless of whether you expect them to come back. Nothing on it is hostile — it is ordinary ownership hygiene that should have been in place from the start.
If the repository turns out to be theirs and inaccessible, the situation is worse but rarely fatal. Whatever is deployed and running still exists, and a competent engineer can usually retrieve a working codebase from the running system.
The database is the part that genuinely cannot be reconstructed, which is why hosting access is more urgent than repository access. Get a backup taken and stored somewhere you control before anything else.
If money is outstanding, it is the only leverage you have, and paying it in the hope of prompting a reply almost never works. Offer it explicitly in exchange for the handover: repository transfer, credentials, and a short written note on how to run the project.
Put that offer in writing, keep it civil, and give a deadline. It converts an emotional standoff into a transaction, which is far more likely to produce a result.
It is almost never malice. The common causes are that the developer took on more than they could deliver and could not face saying so, that the project became unprofitable after they underquoted it, or that they hit a technical problem they could not solve and stopped answering rather than admit it.
That matters practically: it predicts what you will find in the code. Expect the last third of the work to be incomplete rather than the whole thing to be fabricated, and expect one specific hard problem sitting unsolved somewhere near where they stopped.
Secure your accounts now regardless. That step is reversible and costs nothing if they reappear, whereas losing the domain or the database is not reversible at all.
Rarely. Whatever is deployed still exists and can usually be recovered into a working codebase. The database matters more, which is why hosting access is the urgent one.
