# Entry Points

## Goal

Define how a human should think about invoking this skill in OpenClaw.

## Primary entry points

### 1. Start a meeting

Use this when you want multiple roles to discuss one topic.

Suggested spoken intent:
- start a meeting about X
- review X with multiple roles
- ask AI to choose roles for X and then confirm them

### 2. Analyze with one role

Use this when you want a single role to judge a topic.

Suggested spoken intent:
- ask real_powell to evaluate X
- have gstack_technical review X
- get a single role's opinion on X

## What the skill should do first

When the user opens the skill:
1. Accept the topic or analysis target.
2. Ask whether this is meeting mode or single-role mode.
3. If meeting mode, ask for large background text or a source document.
4. Let AI suggest participants.
5. Ask the user to confirm or adjust.

## What the skill should do during the run

- One role turn per message.
- One user interruption per message.
- One PM summary per message when needed.
- One final report at the end.

## Why this matters

OpenClaw skills work best when the first question is simple.
The skill should guide the user from intent to input without forcing them to think about internals.