hakoniwa-business-pack: Instructions file for Codex

Hakoniwa Business Pack Agent Router

This repository is structured system knowledge for translating user requirements into defensible Hakoniwa system compositions and for improving that knowledge from real usage.

Do not treat this repository as a normal source-code project first.

AGENTS.md is the stable entry point for AI agents. It routes the agent to one of two primary roles:

• User / Solution Agent: understand a user goal and produce a defensible Hakoniwa Recipe, with execution or validation only when requested.
• Maintainer / Learning Agent: turn reusable demand signals, implementation discoveries, runtime evidence, and corrections into durable Business Pack knowledge and executable guardrails.

Choose The Active Role

User / Solution Agent

Read AGENTS-USER.md when the task is primarily about using Hakoniwa to satisfy a user need.

Typical triggers:

• "Can Hakoniwa do X?"
• "How should I build X with Hakoniwa?"
• selecting components for a requirement
• designing or updating a Recipe
• evaluating feasibility or validation status
• creating or running a simulation-only Demo
• implementing an already-defined Recipe

The normal flow is:

User Requirement
  -> Search / interpret existing Use Cases
  -> Required Capabilities
  -> Catalog Components
  -> Recipe
  -> Feasibility / Validation / Agency Boundary
  -> Optional Execution
  -> Preserve unmet or reusable demand as Use Case knowledge

Before evaluating feasibility from Catalog components alone, search usecases/ for an existing reusable problem or desired outcome that matches the user's intent. Reuse or refine that Use Case when appropriate.

If no matching Use Case exists, derive a provisional Use Case from the user's goal. A request that is currently not_feasible, partially_feasible, or unknown is still valuable demand knowledge: preserve the unmet need as a Use Case Fragment or canonical Use Case candidate instead of discarding it after feasibility analysis.