Do I need a key?
Yes. You need the web address of your YApi site plus a project token, which you copy from that project's settings page in YApi. The token is written as the project number, a colon, then the token, and you can list several projects separated by commas.
Is setup hard?
No. You paste one short block into your assistant's settings and it fetches the program for you the first time it runs. There is nothing to install by hand.
Which apps does it work in?
Cursor and Claude Desktop are both documented, and any assistant that can start a command should behave the same way.
Can it delete our documentation?
No, there is no delete tool. It can, however, save over an existing entry with no confirmation step, so ask it to show you the change before it writes.
Is my token handled safely?
Mostly. It is masked in the logs, but the setup the README recommends first puts the token on the command line, where anything that can list running programs on your machine can read it. Prefer the version that puts it in the settings block as an environment value instead.
The documentation mentions a silent logging level. Does it work?
No. Asking for the silent level is rejected on the command line and quietly ignored as an environment value, so the program falls back to normal logging. Use the errors-only level if you want it quiet.
Can I use this to document a brand new interface?
Yes. Describe the path, the method and the parameters in plain words and it will create the entry in the category you name.
Will everything be in Chinese?
The project's own documentation and its messages are in Chinese, but you can talk to it in any language and your assistant will translate what comes back.
There is also a local web-server mode. Is it safe?
Only on a machine you trust. That mode listens on every network connection with no password at all, so anyone who can reach your machine could use your YApi token. Stick with the standard setup unless you are debugging.
