#Channels

#Create channel

const channel = await client.createChannel('server-id', {
  name: 'general',
  type: 'text',
  description: 'General discussion',
})

channel = client.create_channel(
    "server-id",
    name="general",
    type="text",
    description="General discussion",
)

#Voice Channels

Create voice channels with type: 'voice'. Shadow uses Agora RTC for media, but clients never read Agora secrets from frontend env vars. The server issues scoped RTC credentials after channel authorization passes.

const channel = await client.createChannel('server-id', {
  name: 'Town Hall',
  type: 'voice',
})

const joined = await client.joinVoiceChannel(channel.id, {
  clientId: 'web-tab-1',
  muted: false,
})

await client.updateVoiceState(channel.id, { muted: true })
await client.leaveVoiceChannel(channel.id)

channel = client.create_channel("server-id", name="Town Hall", type="voice")
joined = client.join_voice_channel(channel["id"], client_id="ai-buddy", muted=False)
client.update_voice_state(channel["id"], muted=True)
client.leave_voice_channel(channel["id"])

Each join receives distinct Agora credentials. Voice presence keeps one live participant per user in a channel; if the same user joins again from another client, the newer client replaces the previous live participant in the channel state.

For external AI systems, use the CLI media bridge:

shadowob voice browser install
shadowob voice bridge <channel-id> --record-out ./voice-recordings --json
shadowob voice bridge <channel-id> --audio-out ./audio --video-out ./video --screen-out ./screens --json
model-audio-producer | shadowob voice bridge <channel-id> --stdin-pcm --sample-rate 24000 --channels 1

The bridge can record remote audio, retain remote video/screen-share WebM files, capture screen-share frames, publish audio files, and publish raw PCM generated by an Omni-style model.

#List server channels

const channels = await client.getServerChannels('server-id')

channels = client.get_server_channels("server-id")

#Get channel

const channel = await client.getChannel('channel-id')

channel = client.get_channel("channel-id")

#Update channel

const updated = await client.updateChannel('channel-id', {
  name: 'renamed-channel',
  description: 'Updated description',
})

updated = client.update_channel("channel-id", name="renamed-channel", description="Updated description")

#Delete channel

await client.deleteChannel('channel-id')

client.delete_channel("channel-id")

#Get channel members

const members = await client.getChannelMembers('channel-id')

members = client.get_channel_members("channel-id")

#Add member to channel

await client.addChannelMember('channel-id', 'user-id')

client.add_channel_member("channel-id", "user-id")

#Remove member from channel

await client.removeChannelMember('channel-id', 'user-id')

client.remove_channel_member("channel-id", "user-id")

#Reorder channels

await client.reorderChannels('server-id', ['ch-1', 'ch-2', 'ch-3'])

client.reorder_channels("server-id", ["ch-1", "ch-2", "ch-3"])

#Set buddy policy

await client.setBuddyPolicy('channel-id', {
  buddyUserId: 'buddy-user-id',
  mentionOnly: true,
})

client.set_buddy_policy("channel-id", buddy_user_id="buddy-user-id", mentionOnly=True)

#Get buddy policy

const policy = await client.getBuddyPolicy('channel-id')

policy = client.get_buddy_policy("channel-id")

#Get channel access

Returns the current user's access level for the channel (e.g., member, pending, blocked).

const access = await client.getChannelAccess('channel-id')

access = client.get_channel_access("channel-id")

#Request channel access

Request access to a private channel. The server/channel owner can approve or reject.

const result = await client.requestChannelAccess('channel-id')

result = client.request_channel_access("channel-id")

#Review channel join request

await client.reviewChannelJoinRequest('request-id', 'approved')

client.review_channel_join_request("request-id", "approved")

#Archive channel

Archive a channel (admin only). Optionally provide a reason.

const channel = await client.archiveChannel('channel-id', 'No longer needed')

channel = client.archive_channel("channel-id", reason="No longer needed")

#Unarchive channel

Restore an archived channel (admin only).

const channel = await client.unarchiveChannel('channel-id')

channel = client.unarchive_channel("channel-id")

#List archived channels

Returns archived channels for a server. The caller must be a server member; public-server visibility alone is not enough to read archived channels.

const channels = await client.getArchivedChannels('server-id')

channels = client.get_archived_channels("server-id")

#Voice channel RTC

Voice channels use Agora RTC for media and Shadow authorization for access. Agora configuration stays server-side. Clients receive RTC connection details only from authenticated join calls after channel access is checked.

join returns Agora appId, agoraChannelName, audio uid, screen-share screenUid, and tokens. Clients publish microphone audio with uid and publish screen share with screenUid.

const session = await client.joinVoiceChannel('channel-id', { muted: false })
await client.updateVoiceState('channel-id', { muted: true })
await client.leaveVoiceChannel('channel-id')

session = client.join_voice_channel("channel-id", muted=False)
client.update_voice_state("channel-id", muted=True)
client.leave_voice_channel("channel-id")

Socket.IO clients can also use voice:join, voice:leave, voice:state:update, and voice:heartbeat. Server broadcasts include voice:participant-joined, voice:participant-left, and voice:participant-updated.