本文へスキップ
TakoAPI
スキル一覧に戻る

ask-agents

teamolab 作

Productivity & Tasks
491ダウンロード数はclawskills.shが公開している値です
0閲覧数
0件のいいねサインインするとこのスキルにいいねできます
—まだ評価がありません
あなたの評価

概要

AI agent for ask agents tasks.

説明

このスキルでできること

Teamo is a senior AI assistant for knowledge workers that handles task breakdown, planning, and delegation. It coordinates specialized sub-agents for research, data analysis, and writing using the McKinsey 7-Step Problem Solving Process. Each step in a complex task is assigned to the sub-agent best suited for that work type.

Rather than executing tasks directly, it acts as an orchestrating layer that assigns each sub-task to the most appropriate specialized agent, keeping costs predictable and preventing the context collapse that comes from a single agent handling everything.

使いどころ

  • Breaking a complex business decision into MECE sub-tasks
  • Delegating deep market research to a specialized research sub-agent
  • Routing data tables and code tasks to a dedicated analyst agent
  • Retrieving internal knowledge base documents for a writing project
  • Producing a structured decision report from multi-source research findings

ワークフロー例

入力

User asks: Analyze the competitive landscape for our product launch and produce a decision report.

エージェント
  1. 1

    Defines the problem across 6 dimensions (description, decision maker, success criteria, constraints, accuracy) and confirms with user

  2. 2

    Calls Divide Agent to MECE-break the problem into two layers of sub-tasks

  3. 3

    Prioritizes sub-tasks by value, feasibility, and dependency order

  4. 4

    Delegates research sub-tasks to Research Team Leader with a structured instruction template covering scope, sources, and must-include points

  5. 5

    Calls chief_editor_desicion with all research reports and logs to produce the final storytelling report

出力

A structured decision report with MECE breakdown, synthesized research findings, and final recommendations saved as a wiki document.

インストール

clawhub install teamolab/ask-agents

CLIコマンドを実行するか、Claude Code、Cursor、Codexなどお使いのコーディングエージェントにプロンプトを貼り付けてください。

リンク