Notes on docs as code.
On storing documentation as Markdown in Git, self-hosting your knowledge base, and owning your content instead of renting it.
- August 26, 2026
Make your knowledge base agent-readable
What agents actually need from a knowledge source — plain text, stable addresses, metadata, a change signal, structure — and how Markdown in Git supplies all five.
Read post → - August 26, 2026
Two writers, one Markdown file: collaboration without a sync engine
No CRDTs, no websockets, no locks. How DocuCommit handles concurrent Markdown edits with a content hash on save and a real three-way Git merge underneath.
Read post → - August 26, 2026
What actually leaves your network when you run DocuCommit
The complete outbound network inventory for DocuCommit: every destination, when it is contacted, what the request contains, and how to opt out.
Read post → - July 10, 2026
How we store docs as Markdown in Git
A look under the hood: how DocuCommit keeps every page as a plain Markdown file, versions it with native Git history, and still gives you search, comments, and diagrams — with no database anywhere.
Read post → - July 10, 2026
Why your docs belong in Git, not a database
Documentation tools trap your knowledge in a database you don't control. Here's the case for storing docs as plain Markdown in your own Git repo — and what you get for free when you do.
Read post →