feat: add core structure and stakeholder analysis skill

[Copilot]

Bootstraps an AI-driven user-research workspace where users converse with an agent (Copilot Chat, Claude, OpenAI, or any LLM) to produce structured research documents in docs/.

Structure

user-research/
├── .github/copilot-instructions.md       # Copilot Chat skill discovery + dispatch
├── docs/README.md                        # Output directory; files written here by agent
└── skills/
    ├── README.md                         # Skill authoring guide
    └── stakeholder-analysis/
        ├── skill.md                      # Tool-agnostic 6-step prompt definition
        ├── output-template.md            # Markdown template the agent fills in
        └── README.md                     # Per-tool invocation instructions

Skills system

Skills are plain Markdown files — no framework dependency, works with any LLM. Each skill defines a Role, Goal, Process (step-by-step conversation flow), and Output Instructions. The agent reads skill.md, conducts the interview one question at a time, then fills in output-template.md and writes the result to docs/.

Stakeholder Analysis skill

Six-step conversational process: project context → identify stakeholders (primary / secondary / key decision-makers with interest, influence, impact, engagement approach) → conflict & alignment check → gaps & risks → review & confirm → produce document.

Output saved as docs/stakeholder-analysis-<project-slug>.md containing a stakeholder register table, influence/impact matrix, conflicts/alliances, risk table, next steps, and revision history.

Multi-tool invocation

Tool	How
GitHub Copilot Chat	"Run the stakeholder analysis skill" — .github/copilot-instructions.md handles dispatch
Claude / ChatGPT / Codex	Paste skill.md as the system prompt
Any LLM	Copy skill.md as the opening instruction

Original prompt

This is going to be a repo for using Claude code or Codex or Copilot chat to do analysis of users and stakeholders and projects. The users of this repo will converse with an agent, and the agent will have sub-agents and skills which are associated with it and help it to do things like:

• Assessing who the stakeholders are
• Making use cases
• Doing personas
• Understanding business needs
• Identifying requirements
• Similar things
  The system will be driven from the AI chat window and will produce a series of docs in the docs directory. What we need to do to set this up is to create a core structure. We're going to focus on a simple skill: interrogating who the stakeholders are (stakeholder analysis). That's the one skill we're going to do first. You're going to set it up so that those skills can be used by either Claude, OpenAI, Codex, or other tools.