Nice try. The moat was never the code. Realistically you won't.

What stops a copy of GitHub working?

Serving git at scale is a distinct and brutal problem: repositories of tens of gigabytes, clone storms from CI, pack file generation under load, and read replicas that must never serve a stale ref. Code search across all public code needs a purpose-built index; general-purpose search engines fall over on the token distribution of source code. Abuse, malware distribution, and being critical infrastructure for supply chains means a security organisation, not a security feature. Availability expectations are absolute, because when it is down a meaningful fraction of the world's engineers stop working.

Can AI copy

GitHub

Git, plus the social layer that made everyone agree on where the code lives.

Nice try

Verdict: Nice try
Time to copy: you won't
Category: ⌘ Dev tools
Source: Proprietary

The prompt

Written to be pasted straight into an agent with no editing. It describes the whole job, including the parts that will go wrong.

Do not build a GitHub. If you want the useful part for yourself, self-host an existing forge — Forgejo or Gitea — and spend your effort on the integrations you actually need. If you want to learn how it works, build a repository browser over a bare git repo: refs, tree navigation, blame, diffs between arbitrary commits, and a pull request model as a merge proposal between two refs with review comments anchored to a blob and line. That teaches you the domain in a week and skips the datacentre.

Open inClaude ↗Terminal agents take it from the clipboard.

What copies cleanly

A repository browser, diff viewer, pull request flow and issue tracker. Genuinely a satisfying month of work.
Git hosting itself, since the hard part was solved by someone else and open-sourced in 2005.
There are several complete open-source implementations, and some are excellent.

Where it stops

Serving git at scale is a distinct and brutal problem: repositories of tens of gigabytes, clone storms from CI, pack file generation under load, and read replicas that must never serve a stale ref.
Code search across all public code needs a purpose-built index; general-purpose search engines fall over on the token distribution of source code.
Abuse, malware distribution, and being critical infrastructure for supply chains means a security organisation, not a security feature.
Availability expectations are absolute, because when it is down a meaningful fraction of the world's engineers stop working.

What actually protects them

Every dependency, CI config, README badge and Stack Overflow answer points at it. The URLs are load-bearing.
Identity. A developer's profile is a portfolio they have spent a decade building and cannot export.
Network effects on contribution: open source lives where the contributors already have accounts.

Editor's note

The interesting thing here is that the copies already exist, are good, and are free — and it has changed almost nothing. That is the clearest evidence available that the product was never the software.

Somebody already did it

Go and look at the original

We would rather you formed your own opinion than took ours. They built the thing; we only measured it.

github.com ↗