When should I trigger this skill?

You should use this skill at the very beginning of a project when you are asked to 'Create a new agent' or 'Start an interview' for a specific domain-specific task.

How does the QC process work within this skill?

It delegates quality control to a specialized sub-agent (qa-skill-qc) that evaluates the documentation for ambiguity, missing items, and alignment with industry standards before finalizing.

What happens if the requirements are not approved?

The workflow includes a mandatory structure approval step; the skill will not proceed to the backlog or next engineering phases until the user approves the agent requirements and folder design.

What is the main output of the Agent Discovery skill?

The skill generates standardized requirements (draft_requirements.md) and industry research documents (framework_research.md) to serve as a technical blueprint for new AI agents.

Does it support industry-standard frameworks?

Yes, the skill is designed to search for and incorporate standards like BABOK or AMA to ensure the resulting agent follows professional domain-specific best practices.

Agent Discovery & Requirements

Name: Agent Discovery & Requirements
Author: matsuni-kk

bymatsuni-kk

•

生産性とワークフロー

Orchestrates the discovery phase for custom AI agents by conducting stakeholder interviews, researching industry frameworks, and generating structured requirement documents.

This skill automates the initial requirement-gathering phase for creating domain-specific Claude agents. It guides users through a comprehensive discovery workflow, including industry-standard research (like BABOK or AMA), detailed persona and KPI definition, and automated quality control loops using specialized sub-agents. By generating standardized documents like draft_requirements.md and framework_research.md, it ensures that every new agent project starts with a solid, peer-reviewed architectural foundation and clear success criteria, streamlining the transition from concept to implementation.

主な機能

01Integrated quality control (QC) loops using dedicated sub-agents

02Template-based generation of requirements and research documentation

03Automated task backlog generation to track project next steps

04Automated industry framework research (BABOK, AMA) via web search

05Structured stakeholder interview workflow for persona and KPI mapping

068 GitHub stars

ユースケース

01Initiating a new specialized AI agent project with formal requirement gathering

02Standardizing the documentation process for enterprise-grade AI agent development

03Researching domain-specific best practices before developing automation workflows

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add matsuni-kk/agent_template_public agent-discovery

For use in Claude.ai and ChatGPT

Download Skill