| name | dingtalk |
| description | DingTalk channel skill. When $dingtalk appears in message context, return your response as text
and the framework will deliver it. For programmatic sends (e.g. progress updates), use dingtalk_send.
|
| metadata | {"channel":"dingtalk"} |
DingTalk Skill
When the message context contains $dingtalk and chat_id, you are in a DingTalk conversation.
Primary: Return Your Response Directly
To reply to a DingTalk user, simply return your response as text. The framework will automatically deliver it to the correct conversation. No script call is required.
- The
chat_id from the message context is used automatically.
- Just write your answer and end the turn. The user will receive it.
Example: User asks "What is 2+2?" → You return "4" → The framework sends "4" to DingTalk.
When to Use the Script
Use dingtalk_send only when you need to send from within a tool (e.g. a progress update during a long-running task, or a status message triggered by another tool).
uv run ./scripts/dingtalk_send.py --chat-id <CHAT_ID> --content "<TEXT>"
Get chat_id from the message context JSON (the chat_id field). For group chats it is group:<openConversationId>, for 1:1 it is the user's staff_id.
Context Fields
When $dingtalk is in scope, the message context typically includes:
channel: $dingtalk — identifies this as a DingTalk conversationchat_id: target for sending (e.g. 204818006723348842 for 1:1, or group:xxx for groups)
Response Policy
- Direct answers: Return your response. The framework delivers it.
- Long-running tasks: Return a short acknowledgment first (e.g. "Working on it..."), then continue. When done, return the final result.
- Errors: Return an error message so the user sees it immediately.
Script Reference (for programmatic use)
uv run ./scripts/dingtalk_send.py \
--chat-id <CHAT_ID> \
--content "<TEXT>" \
[--title "<TITLE>"]
--chat-id, -c: required--content, -m: required--title, -t: optional, default "Bub Reply"
