What this is, and how it pays for itself

What this site is

Code Licenses answers the questions developers actually type when they are trying to work out what a license permits, who owns the code they wrote, and whether the repositories they stopped opening years ago are worth anything. It is written for people who will check the reasoning, so the claims here stay narrow and the numbers stay on the two pages that can support them.

How code licensing works

AI labs train coding models on source code. The public supply has been collected already and it is not representative: it leans heavily towards libraries, frameworks, and the kinds of projects people choose to publish. Most software in the world is not that. It is line-of-business systems, integrations, and long-lived internal tools that were never published.

Synthetic code turns out to be a poor substitute. A model trained on generated examples learns the shape of clean code and not much about what happens when a requirement changes three years in. Real repositories with real commit history carry that, which is why there is a market for them at all. Where training data actually comes from is in where does AI training data come from.

The licences are non-exclusive. Nothing is transferred and nothing is published. You keep the copyright and keep using the code. And because they are non-exclusive, the same repository can be licensed again to a different buyer, with the owner paid each time rather than once. If you have been reading "sell your code" as "hand over your code", that is the misunderstanding to drop.

What makes a repository eligible

┌─ ELIGIBILITY ───────────────────────────────────────────────────────┐
│                                                                     │
│  1. Ownership     you hold the IP, not a client                     │
│  2. Privacy       never published, never forked from public         │
│  3. Substance     real production code, history intact              │
│  4. History       unsquashed. It is most of the value.              │
│  5. Dependencies  vendored and generated code do not count          │
│                                                                     │
│  Fail the first two and the repo is out entirely.                   │
└─────────────────────────────────────────────────────────────────────┘

Ownership is the one that catches people. If the repository came out of client work, what you can do with it was decided by a contract you signed, possibly years ago. Client-owned code, anything under an NDA, and anything holding confidential data stays out of it. That leaves your own products, internal tools, R and D, and engagements that wound down with rights retained, which for most people is more than they expect.

Cleaning up is not your job and you should not do it. Secrets and personal data are stripped from the review copy, in the code and in the history, and verified before a buyer sees anything. Your own repository is never modified. Tidying a repository yourself, especially squashing history, destroys value rather than adding it.

How this site makes money

Plainly: every button on the site leads to a short form on this site. The request goes to the company that does the assessment and licensing, which registers it as a request from this site. You then sign up on its licensing platform via the button on the success screen, sign the agreement and connect GitHub. If a licensing deal happens for someone who came through that form, this site is paid. That is the entire business model. There is nothing to buy here and nothing is gated.

What that does not change: what gets recommended, what is described as eligible, or what is said about the risks. The site is paid only when a deal happens, and a request from someone whose repositories do not qualify is worth nothing to anyone. Being accurate about what does not work is the commercially sensible position as well as the honest one, which is a convenient alignment and you should still hold the site to it.

The buttons are plain links to the form page, in the same tab. The only link that leaves the site is the signup button after the form is sent, and it passes no ranking signal. What the form collects and where it goes is on the privacy page.

Who runs this

A developer with an agency background and about twenty years of shipping client code. The idea for the site came from noticing a personal archive of finished, private, paid-for projects sitting in cold storage, and realising there was now a reason to care about them.

No newsletter, no personal brand, no consulting funnel behind this. If something here is wrong, it should be corrected rather than defended.

What this site is not

It is not legal advice. Contract and copyright questions turn on the specific wording of what you signed and on the jurisdiction you signed it in. The guides here will help you understand what to look for and what the terms mean. They cannot tell you what your contract does, and anyone claiming otherwise from a web page is guessing.

It is also not a marketplace. The assessment, the terms, and the payment all happen on the licensing platform. This site explains the landscape and points at the route.