If you’re looking to use this repo to retrieve video or audio streams from meeting platforms like Zoom, Google Meet, Microsoft Teams, consider checking out Recall.ai, an API for meeting recording.

Good bot support is part of what makes Discord so nice to use. Unfortunately, the official Zoom API is basically only useful for scheduling meetings and using Zoom Chat, not for making in-meeting bots. So I decided to bring this feature to Zoom myself.

Here is a demo of a basic, ~130 lines of code bot:

I created this by reverse engineering the Zoom Web SDK. Regular web joins are captcha-gated but web SDK joins are not. I use an API only used by the Web SDK to get tokens needed to join the meeting. This means you need a Zoom API key/secret, specifically a "Meeting SDK" one. These can be obtained on the Zoom App Marketplace site: click Meeting SDK (Create) -> name app, disable publishing to marketplace -> fill descriptions and contact information with anything you want -> click App Credentials. The demo at cmd/zoomer/main.go reads these from the environment as ZOOM_API_KEY and ZOOM_API_SECRET.

Because the API keys are associated with your account, using this software may get your Zoom account banned (reverse engineering is against the Zoom Terms of Service). Please do not use this on an important account.

$ go get github.com/chris124567/zoomer
$ cd $GOPATH/src/github.com/chris124567/zoomer
$ scripts/build.sh
$ ZOOM_API_KEY="xxx" ZOOM_API_SECRET="xxx" ./zoomer -meetingNumber xxxxx -password xxxxx

Feel free to use the demo as a template. If you want to use the library elsewhere just import github.com/chris124567/zoomer/pkg/zoom.

See the comments in cmd/zoomer/main.go

Note that you are free to construct your own message types for any I have not implemented.

For sending: Look at zoom/requests.go and switch out the struct and message type names for your new message type

For receiving: Create a definition for the type and update the getPointerForBody function in zoom/message.go.

The protocol used by the Zoom Web client is basically just JSON over Websockets. The messages look something like this:

{"body":{"bCanUnmuteVideo":true},"evt":7938,"seq":44}
{"body":{"add":null,"remove":null,"update":[{"audio":"","bAudioUnencrytped":false,"id":16785408}]},"evt":7937,"seq":47}
{"body":{"add":null,"remove":null,"update":[{"caps":5,"id":16785408,"muted":true}]},"evt":7937,"seq":63}
{"body":{"dc":"the United States(SC)","network":"Zoom Global Network","region":"the United States"},"evt":7954,"seq":3}

The "evt" number specifies the event number. There is a (mostly complete) list of these in zoom/constant.go that I extracted from javascript code on the meeting page.

For the above three messages, the types are:

WS_CONF_ATTRIBUTE_INDICATION                     = 7938 // ConferenceAttributeIndication
WS_CONF_ROSTER_INDICATION                        = 7937 // ConferenceRosterIndication
WS_CONF_DC_REGION_INDICATION                     = 7954 // ConferenceDCRegionIndication

The thing in the comments to the right is the struct type for that message, which can be found in zoom/message_types.go.

Also, the server and client both have sequence numbers ("seq") for the messages they send but it doesn't appear to be used for anything (?).

• Gracefully exit/disconnect
• Organize zoom/message_types.go and general refactoring
• Support for meetings where you don't have the password but just a Zoom url with the "pwd" parameter in it (anyone know anything about this??)
• Thoroughly test things
• Make it more extensible
• Joining breakout room support
• More comments and documentation
• Support audio/video

This is hobbyist software that has no guarantees of being maintained or supported. Please don't use it anywhere near production.