Documentation people can trust
Docs are how people — and now AI tools — learn your product. We’re building the platform that keeps them accurate.
Why Docsheaf exists
Most documentation starts out right. Then the product changes: a flag is renamed, a default moves, a screen gets redesigned. The page that was correct last quarter is now quietly wrong, and nobody finds out until a reader does.
That used to cost a support ticket. Today it costs more, because assistants and coding agents read your docs too, and they repeat what they find with complete confidence.
Software teams solved a version of this problem long ago with review: every change goes through a branch, a diff and a second pair of eyes. Docsheaf brings that discipline to documentation without asking writers to learn Git — a block editor, change requests that merge block by block, and published sites that tell you what readers searched for and couldn’t find.
Principles
What we believe about docs
These shape every feature we build — and the ones we decide not to.
Accuracy over volume
More pages don’t help if readers can’t trust them. We build for docs that stay correct, not docs that pile up.
Review is a feature
Changes to docs deserve the same care as changes to code: a branch, a diff, a reviewer and a clean merge.
Open formats, no lock-in
Markdown goes in and Markdown comes out. Every published page has a Markdown twin, and everything is in the API.
Readers first
Fast pages, instant search, clear navigation and a way to say a page didn’t help.
Secure by default
Roles on every route, two-factor sign-in, sessions you can see and revoke, and an audit trail.
Your infrastructure, your choice
Docsheaf runs on standard Postgres and S3-compatible storage, so your content lives where you decide.
Where we are
What Docsheaf does today
- A block editor with hints, tabs, steppers, code, tables and embeds
- Change requests with reviews, diffs, three-way merge and “update from main”
- Docs sites with sections, themes, custom domains and unlisted share links
- Search, page feedback and analytics that surface content gaps
- llms.txt, llms-full.txt and a Markdown version of every published page
- A REST API with personal tokens, two-factor sign-in and audit logs
Real-time co-editing, Git Sync with GitLab and an MCP server for every site are next. The changelog tracks what ships.
Want to help build it?
We’d love to hear from people who care about documentation — as customers, contributors or colleagues.