CtrlK
BlogDocsLog inGet started
Tessl Logo

communication

Use before sending @mentions, deciding whether to reply, TASK_COMPLETED, BLOCKED, QUESTION, direct answers, loop prevention, or suppressing low-information acknowledgements.

74

Quality

91%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Communication

Routing

Always reply directly in the current room. Do not use the message tool for cross-room sends — Workers do not communicate across rooms.

Coordinator Identity

Your coordinator is the sender of the current task assignment message. Use that Matrix ID directly for @mentions. Do not call organization or agt CLI to look up your coordinator during standard task flows.

@Mention Rules

Use a full Matrix ID when the recipient must act.

Mention your coordinator only for:

  • Task completion: @coordinator:domain TASK_COMPLETED: <summary>
  • Blocker: @coordinator:domain BLOCKED: <what is blocking you>
  • Question: @coordinator:domain QUESTION: <your question>
  • Direct answer to a coordinator question

Do not @mention for:

  • "Got it"
  • "Thanks"
  • "Working on it"
  • Encouragement-only replies
  • Status symbols such as green dots or check marks
  • Short acknowledgments such as ok, done, 收到, or 好的
  • Mid-task progress that requires no decision

Exception: when a new assigned task arrives, task-management requires you to directly say in the current room that you received the message before task acceptance work starts. Do not turn it into a progress thread, and do not send repeated acknowledgements for the same task.

Before sending any @mention, remove all Matrix IDs from the message in your head. Send only if the remaining text contains a concrete completion, blocker, question, requested answer, or decision.

History Context

If your message includes a history section, treat it as context only. Act on the current message section.

Loop Safeguard

If two rounds of replies produce no new task, question, or decision, stop replying.

Repository
agentscope-ai/AgentTeams
Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.