What you want to know beforehand, answered honestly.

Nothing. nexbrand is free software under the AGPL-3.0. There is no paid tier, and you need no account with anyone, only a machine that runs Docker.

For agencies, freelancers and developers who work for several clients and need their colours, fonts, logos and rules in one place, with everything that applied over the years. If you look after a single brand only, it works too, but you use only part of it.

No. nexbrand is a workbench for colours, fonts and sizes, not a drawing program. You draw logos in the program you are used to and upload them. Until there is one, nexbrand sets word marks from name, font and main colour as stand-ins.

No, not one of its own. The export as design tokens (JSON) follows the W3C draft, and tools that read that format can work with it. A connection that sends changes back and forth does not exist.

German and English. Every further language is a JSON file the operator uploads under Settings, Server, Languages; a template can be downloaded in the same place. Whatever such a file lacks, nexbrand shows in English.

nexbrand asks GitHub once a day whether there is a newer version and shows it under “About nexbrand”. After an update it explains to every account, once, what is new. How to update is described under Self-host.

Create the client and choose “Begin empty”, or let nexbrand collect from the website, the repository or the PDF style guide. When publishing you give the date the guidelines applied from back then. You add older versions the same way, each with its own date.

No, on purpose. A published version stays as it was, so everybody can rely on it. For a correction you create a new version that starts with the old one, change what was wrong, and publish it.

Up to 48 colours, 12 fonts, 24 logos, 12 type sizes, 60 rules and 60 examples. A client can have up to 300 versions, 30 drafts and 300 projects. A single file may be 50 MB by default, and the operator can lower that.

No. An uploaded file carries its own colours; nexbrand shows it on its ground but does not check it. Drawn word marks, whose colours come from the palette, are checked, and so are rules that tie a logo to certain grounds.

A client stays in the trash for 30 days and comes back with “Restore”, with its versions and projects. Published versions cannot be deleted at all, only drafts. Only someone who manages the client can delete it for good, and only with their own password.

nexbrand runs in the browser, so on a phone too. For editing brand guidelines, though, a larger screen is much more comfortable. There is no app in a store.

No. For a review and for a public page they get a link. If they want to work in nexbrand themselves, you invite them and give them access to exactly their client, to read, write or manage.

No. A client somebody has no access to does not exist for them, not even its name. Only the operator sees and manages all of them.

Only if you say so: when sending a review you can give an address, and with a mail server set up nexbrand mails the link. Otherwise nexbrand sends mail only for invitations and the operator's test mail, no notifications.

No. A font with the licence “In-house only” appears on public pages, in reviews, through the API and over MCP by its name only, never as a file. Whoever needs the file gets it the way the licence allows.

On your server, in nexbrand's data folder: the database and next to it every logo, font and example. There is no service of ours that anything is sent to.

On its own, only one request a day to GitHub asking whether there is a new version, without names or content; the operator can switch it off. Everything else happens only when someone triggers it and the operator has allowed it: collecting from a website or a repository, a request to your own AI service, a mail through your mail server, signing in through your provider.

Only if the operator allows it and an account enters a service of its own, and then only for two tasks: sorting what was collected and proposing directions. Every request is listed in that account word for word for 14 days. With Ollama on a machine of your own, nothing leaves your network.

Secrets on the server, such as the mail server's password, the keys of AI services and the list of what went to the AI, are stored encrypted. The database itself and the backups are not; protect the data folder and downloaded backups like anything confidential. On the way, the reverse proxy encrypts with HTTPS.

No. The log never contains designs, passwords, keys or tokens. It says who signed in, what went wrong and, when collecting, who confirmed the licence of a font.

A machine with Docker, for Intel and AMD as well as for ARM, so a Raspberry Pi or a NAS works too. nexbrand runs as one container with an SQLite database; you need no database server of your own. For reviews and public pages a reverse proxy with HTTPS goes in front. How that works is described under Self-host.

No. SQLite does not cope with shares over SMB or NFS. Put /data on a disk attached directly to the machine.

With its own backups under Settings, Server, Backups. By default one is made every night, and the last seven are kept. A backup is a ZIP with the database, every file and the key. “Check” shows whether it is complete and what a restore would change. Keep copies somewhere else as well.

Yes, over OpenID Connect. For authentik there is a button that creates the provider and the application itself, and a blueprint file if you prefer. Other services you enter with address, client ID and secret.

Yes. Programs read clients, versions and code with a token and check colour values against the guidelines, AI tools in your editor use MCP. A token of the level “Read and propose drafts” writes drafts back but cannot publish. More under For programs.