worldmaker

name	worldmaker
description	Launch an AI world designer to test the NeoMud Maker editor through browser interaction
context	fork
agent	worldmaker

Test NeoMud Maker Editor

Launch a WorldMaker session using the Playwright MCP browser tools. The worldmaker agent will open a real browser, navigate the Maker editor, take screenshots, and provide feedback from a world designer's perspective.

Instructions

$ARGUMENTS

If arguments were provided above, focus your testing session on that specific area (e.g., "zone editor", "item CRUD", "NPC creation", "export flow"). Explore naturally but concentrate your effort and feedback on the requested system.

If no arguments were provided, run an exploratory session: navigate every section, try creating and editing entities, test the zone map editor, check visual polish — evaluate the editor as a new world designer would.

Prerequisites

First, check if the Maker dev server is running:

curl -s http://localhost:5173 > /dev/null 2>&1 && echo "Maker is running" || echo "Maker is NOT running"

If the Maker is not running, inform the user that it needs to be started:

The Maker dev server is not running. Please start it first:
  cd maker && npm run dev

Do NOT proceed until the Maker is confirmed running.

Session Flow

1. Navigate to http://localhost:5173 using browser_navigate
2. Take an initial screenshot to see the landing page
3. Follow your methodology — look, understand, assess, decide, act, evaluate, report
4. File GitHub issues for every bug and UX problem you encounter
5. End with a structured WorldMaker report

Cleanup (MANDATORY)

After the session is complete (after producing the report), you MUST run these cleanup commands. Do NOT skip this step:

# Delete all screenshots and world copies created during the session
rm -rf .playwright-mcp/*

This removes screenshots (page-*.png) and any world bundle copies (.nmd files) that Playwright cached during the session. These files are large and must not accumulate.