Skip to main content

Channels and Voice

Anda can stay in the terminal, or it can join team channels. Channels bring context back into the same runtime and memory system.

Supported Channels​

  • IRC
  • Telegram
  • WeChat
  • Discord
  • Lark / Feishu

Channel configuration lives under channels in ~/.anda/config.yaml.

Telegram Example​

channels:
  telegram:
    - id: personal
      bot_token: "YOUR_TELEGRAM_BOT_TOKEN"
      username: "YOUR_TELEGRAM_BOT_USERNAME"
      allowed_users:
        - "*"
      mention_only: false

WeChat Example​

channels:
  wechat:
    - id: personal
      # Leave empty to run anda channel init wechat, scan the QR code, and get a token.
      bot_token: ""
      username: anda-wechat
      allowed_users:
        - "*"

Lark / Feishu Example​

channels:
  lark:
    - id: work
      app_id: "cli_xxx"
      app_secret: "YOUR_APP_SECRET"
      platform: feishu
      receive_mode: websocket
      mention_only: true

Channel Safety Checks​

  • allowed_users decides who can trigger Anda; use "*" only when open access is acceptable.
  • Telegram and Discord usually require bot_token.
  • WeChat can use a saved token, or QR login when bot_token is empty.
  • Lark / Feishu needs app_id and app_secret; Feishu endpoints use platform: feishu.
  • Channel routes are persisted so later replies can return to the original thread, room, or contact.

Voice Input and Playback​

Voice input requires transcription.enabled: true. Start a voice session with:

anda voice --record-secs 8

If you also want Anda to play the answer, configure tts.enabled: true. For microphone input and text output only:

anda voice --record-secs 8 --no-playback

Audio attachments can enter context after transcription is enabled, so voice messages in channels can become part of the same memory thread.