Feishu / Lark

The Feishu driver receives messages via an HTTP webhook pushed by Feishu's event system and sends via the Feishu IM v1 API using lark-oapi.

Feishu (China) and Lark (international) use the same API and the same driver.

Setup

1. Go to the Feishu Open Platform (or Lark Developer).
2. Create a custom app and enable the im:message:receive_v1 event.
3. Under Event Subscriptions, set the request URL to http://your-host:8080/event (matching your listen_port and listen_path).
4. Copy the App ID, App Secret, Verification Token, and Encrypt Key (leave encrypt key blank to disable encryption).
5. Add the bot to the target group chat.

Public endpoint required

Feishu must be able to reach your HTTP endpoint from the internet. Use a reverse proxy, tunnel (e.g. ngrok), or deploy on a public server.

Config keys

Add under feishu.<instance_id> in config.json:

Key	Required	Default	Description
app_id	Yes	—	Feishu/Lark App ID
app_secret	Yes	—	Feishu/Lark App Secret
verification_token	No	""	Event verification token from the developer console
encrypt_key	No	""	Event encryption key (leave empty to disable)
listen_port	No	8080	HTTP port to listen on for incoming events
listen_path	No	"/event"	HTTP path for incoming events

{
  "feishu": {
    "fs_main": {
      "app_id": "cli_xxxxxxxxxxxx",
      "app_secret": "your_app_secret",
      "verification_token": "your_verification_token",
      "encrypt_key": "",
      "listen_port": 8080,
      "listen_path": "/event"
    }
  }
}

Rule channel keys

Use under channels or from/to in rules.json:

Key	Description
chat_id	Feishu open chat ID, e.g. "oc_xxxxxxxxxxxxxxxxxx"

{
  "fs_main": { "chat_id": "oc_xxxxxxxxxxxxxxxxxx" }
}

You can find the chat ID in the Feishu developer console, or from the event payload of any message sent to the bot in that chat.

Notes

• Currently only text messages are received. Other message types (cards, files, stickers) are ignored on the receive side.
• Outgoing attachments are sent as URLs appended to the message text (Feishu file upload via the API requires additional permissions).
• The sender's display name is shown as their open_id. Resolving the human-readable name requires an extra user-info API call and is not currently implemented.