IM bots (Lark / DingTalk / Telegram)

Niuniu’s IM bots let you drive agents straight from a Lark / DingTalk / Telegram group — without opening the desktop app: describe the work in a group, Niuniu spins up a workspace in the matching project, and the agent’s progress and results come back conversationally in the chat.

The core idea in one line: one bot (one credential, one connection) routes different groups to different projects under your account.

Supported platforms

Platform	Connection	Inbound	Outbound
Lark / 飞书	Long-lived connection (no public callback needed)	Text, images, attachments	Markdown, receipts, task-status emoji
DingTalk	Long-lived connection	Text, images, attachments	Markdown, receipts, task-status emoji
Telegram	Bot API	Text, images, attachments	Markdown, receipts, new-reply

Step 1 — Create a bot

Go to Settings → IM bots → + New bot.

1. Pick a platform (Lark / DingTalk / Telegram) and enter the bot credentials issued by that platform’s developer console (App ID / Secret or Bot Token, per the field hints).
2. After saving, the bot comes online over a long-lived connection and shows Enabled; hit Test anytime to verify connectivity.

One bot can serve many groups and many projects — you don’t need a separate bot per project.

Step 2 — Bind groups to projects

Once the bot is in a group, tell Niuniu which project that group’s messages should land in.

• In the bot’s pending chats list, select a group, choose a target project, and confirm the binding.
• On success, the page shows that group’s channel_id — handy for reference and troubleshooting.
• Multi-project routing: bind different groups to different projects, and a single bot covers all your projects; group sessions are shown per their routed target project, so nothing gets crossed.

Bindings are managed centrally under Configured bindings: review every chat ↔ project mapping and unbind at any time.

Step 3 — Collaborate conversationally in the group

Once bound, the group becomes an entry point:

• Ask for work: @-mention the bot (or message per platform rules), and Niuniu spins up a workspace in the matching project for the agent to execute.
• Two-way conversation: the agent’s real replies come back to the group; you can follow up and course-correct in a back-and-forth, not a fire-and-forget.
• Images in: drop screenshots or log files right into the chat and the agent reads them too.
• Markdown out: the agent’s replies render as markdown — code blocks and lists stay readable.
• Session switching: a group may have several sessions in flight; switch by session ID to attach a reply to the right context.
• Receipts & task status: messages get delivery receipts, and in-progress work is shown with each platform’s native emoji reactions.

Typical scenarios

• Quick tickets: in a dev group, @-mention the bot with “look into the login-page error log”; Niuniu spins up a workspace in the matching project and the agent posts its findings back.
• On-call response: bind an alerts group to your ops project, and when an alert fires, have the agent pull logs and give a first triage right in the chat.
• One bot, many projects: a single bot serves the “frontend”, “backend”, and “ops” groups at once, each routed to its own project without interference.
• Drive from mobile: on your phone, hand work to the agent and check progress via Lark / DingTalk, then dig in later at your desk.

Credentials & binding management

• View bindings: Settings → IM bots → Configured bindings lists each chat ↔ project pair.
• Unbind: delete a binding from that list, and the group’s messages stop landing in the old project.
• channel_id: shown after creating / pairing, it’s the group’s unique identifier for troubleshooting routing.

Troubleshooting

Symptom	Cause / fix
Bot offline	Wrong credentials or not enabled on the platform side — re-enter credentials and hit Test
No response to @-mention	The group isn’t bound to a project — check Configured bindings or re-pair
Messages land in the wrong project	Group bound to the wrong project — unbind and rebind to the right one
Images / attachments not read	Make sure the platform bot has permission to read message attachments

Next steps

• Kanban & Issues — how a request raised in chat flows once it becomes an issue
• Workspace & Agent chat — how the agent works inside a workspace
• External integrations — have the agent query GitHub / Jira / TAPD