Every docs tool can host an article. None of them know when a merged PR just made it wrong. Ellie Docs connects to your GitHub repo, tracks the code each article actually depends on, and flags — then drafts — the fix the moment it goes stale.
Early access · no spam, ever · we'll email you the moment it's ready
A field gets renamed in the API. An endpoint gets deprecated. A UI flow gets redesigned. The PR merges clean, tests pass, and the help article referencing the old behaviour sits there — correct-looking, silently wrong — until a customer or a support ticket catches it.
Each article can be tied to the code, endpoints, or API spec it actually documents — so Ellie Docs knows exactly what to watch.
A GitHub App watches merged PRs, judges whether they affect a linked article, and flags it — with a drafted revision ready for review, not just a red banner.
Have an OpenAPI spec? Ellie keeps the reference in sync with it, diffed for breaking changes on every version. No spec? It derives one from your routes and request validators, citing the code behind each field.
Install the Ellie Docs GitHub App on the repos your product docs depend on. We don't retain a copy of your repository — code is fetched when Ellie needs to read it and re-fetched by commit later.
Bring existing docs in via crawl-import, or start fresh — Ellie Docs drafts articles and links each one to the code, endpoints, or spec it depends on.
When a merge affects a linked article, Ellie flags it and drafts the revision. Your team reviews a queue of proposed fixes instead of hunting for what broke.
A GitHub App scoped to the repos you choose. We don't retain a copy of your repository — Ellie fetches the tree when it needs to read it and re-fetches by commit afterwards. What is kept is the derived material an article stands on: the excerpts and citations behind each claim, so any statement in your docs can be traced back to the code that justifies it. Secrets found in files are redacted before storage, and your code is never used to train models.
Yes, that's the intent — Ellie Docs hosts and renders the public help centre directly, including custom domains, so you're not running two systems in parallel.
Yes. Point Ellie Docs at a spec and it maintains the reference from it, diffed against each new version as your API evolves. Without one, Ellie derives the spec from your code — endpoints from your router, field types from your request validators, and for anything the code doesn't state outright, an investigation that has to cite the lines it drew each field from. Derived endpoints land private until you review them, so internal routes don't leak into your public reference.
Pair Ellie Docs with Ellie Desk (also in early access): resolved support tickets can become new doc articles in one click, and Desk's AI drafts get grounded in your published docs automatically.
