genesis-worldbuilder

name	genesis-worldbuilder
description	Generate a complete living world from a natural language concept — geography, factions, characters with genetic traits
version	1.0.0
metadata	{"hermes":{"tags":["worldbuilding","procedural-generation","fiction","creative","genesis"],"related_skills":["genesis-simulator"],"category":"creative"}}

Genesis Worldbuilder

Generate complete living worlds from natural language descriptions. Creates geography, factions, and characters with genetic trait systems.

When to Use

When the user asks to:

• Create a world, universe, or fictional setting
• Generate a fantasy/sci-fi/post-apocalyptic world
• Build a setting for a story, game, or campaign
• "Make me a world about..."

Prerequisites

• Genesis API running (FastAPI backend at configured URL)
• Nous Research API key configured

Procedure

Phase 1: Parse the Concept

Extract from the user's description:

• Theme (fantasy, sci-fi, post-apocalyptic, historical, etc.)
• Core conflict (resource scarcity, political rivalry, ideological war, etc.)
• Scale (village, kingdom, continent, planet, galaxy)
• Tone (dark, hopeful, gritty, whimsical)

If the description is vague, ask ONE clarifying question. Don't over-ask — generate and refine.

Phase 2: Generate World Elements

Use delegate_task with batch mode (3 parallel tasks):

Task 1 — Geography:

goal: "Generate 6 regions for a world with this concept: [SEED]. Output JSON with regions array (id, name, type, climate, resources, neighbors, x, y, description) and connections array."
toolsets: ["web"]

Task 2 — Factions:

goal: "Generate 4 factions for a world with this concept: [SEED] and these regions: [REGION_SUMMARY]. Output JSON array with id, name, ideology, color, territory, resources, alliances, enemies, population, morale, traits, description."
toolsets: ["web"]

Task 3 — Characters:

goal: "Generate 15 characters for a world with this concept: [SEED], these factions: [FACTION_SUMMARY], and these regions: [REGION_SUMMARY]. Each character needs id, name, faction_id, role, age, location, backstory, goals, relationships, and genome (courage/cunning/loyalty/ambition/empathy/resilience as 0-1 values). Output JSON array."
toolsets: ["web"]

Phase 3: Assemble and Deploy

1. Combine geography, factions, and characters into a world JSON
2. POST to Genesis API: /api/worlds with the seed
3. Report the live URL to the user
4. Offer to simulate days or explore the world

Phase 4: Schedule Autonomous Simulation (Optional)

If the user wants the world to evolve on its own, use schedule_cronjob:

prompt: "Load world [WORLD_ID] from Genesis API at [API_URL]. Call POST /api/worlds/[WORLD_ID]/simulate?days=1 to advance one day. Then GET the latest events and summarize the key developments in 2-3 sentences."
schedule: "every 1h"
deliver: "telegram"
name: "genesis-[WORLD_NAME]-sim"

Example Interaction

User: "Create a world where magic is dying and three noble houses fight over the last wellspring"

Agent:

1. Parses: fantasy theme, resource conflict (magic wellspring), kingdom scale, dramatic tone
2. Delegates 3 parallel tasks for geography/factions/characters
3. Assembles world, posts to API
4. Returns: "Your world 'The Fading Realm' is live at [URL]. 6 regions, 3 noble houses, 15 characters. Want me to start simulating events?"

Pitfalls

• Don't generate more than 20 characters — quality drops
• Ensure every region is controlled by exactly one faction
• Genome values must be 0.0-1.0
• Character relationships must reference valid character IDs
• If LLM returns invalid JSON, retry once with a stricter prompt