Open Source Document Management

People look for open source document management for reasons that are worth separating, because they lead to different choices. Some want to avoid a per-seat subscription. Some cannot put documents on somebody else's infrastructure. Some want to know they can still open their archive in fifteen years. Only the second and third are really about open source; the first is about price and is often answered by a free tier instead.

We sell source code you clone and run yourself, so we are in the same broad family, though not a document product. This page is about the category first.

The reasons, and what each actually implies

What to evaluate, beyond the feature list

Open source document systems vary more in operational character than in features, and the features are rarely what causes regret.

Storage growth is the cost nobody models

Documents accumulate, versions multiply them, and previews and extracted text add more. An archive that starts at fifty gigabytes is not going to stay there, and the backup of it grows at the same rate.

Before choosing, work out roughly how much you add per year and multiply by the number of years you are required to retain. That number, rather than the licence, is usually what determines whether self-hosting is sensible.

When you would build rather than install

Installing an existing system is nearly always right for a general repository, because the boring parts, OCR, previews, versioning, permissions, are done and tested.

The case for building is documents that belong to a process rather than to a library. If every document arrives attached to a case, a claim, a client matter, or a compliance requirement, then a general repository is the wrong shape and integrating one is often more work than storing files against the record they belong to.

What you would start from here

There is no document management in this stack, and no storage, versioning, OCR, previews, or search over document contents.

What the Vault template provides is the application documents would attach to: organizations as the boundary, roles and invitations so access is scoped, a Hono backend with Prisma and Kysely on Postgres, and an admin app. Since the whole point of this search is usually self-hosting, it is worth being clear that this is source you run on your own infrastructure, with no per-seat pricing between you and your users, which is the same property people come to open source document systems for.

You would write the document and version records, put the files in object storage with the database holding the pointer, enforce permission checks at the point the file is served rather than only in the interface, and add extraction or preview generation as background work if you need it. Postgres full text search will carry you further than expected before a dedicated search service is justified.

If you need a general repository, install one of the established projects. The packs below are for documents that belong to a process you are building.