thesethrose/dungeon-agent/build-dungeon

Claude Code and agent skill by thesethrose/dungeon-agent

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Process skill: design a dungeon / map / level layout collaboratively for the current campaign. Grills the player (Hermes clarify/Ask-Question, one question at a time) about size, theme, difficulty curve, branching, hazards, and boss — then produces a dungeon layout the DM runs. Run when the player says 'make me a dungeon', 'design the layout', or 'create a level'. Depends on the world's monster/loot pool.

Install

npx vskill@latest install thesethrose/dungeon-agent/build-dungeon

SKILL.md

Build a Dungeon (layout design, grilling-driven)

A process skill. Designs a specific dungeon / map / level for the current campaign by interviewing the player one decision at a time (see grilling), then produces a layout the DM uses to run it. It pulls monsters, loot, and items from the campaign's world.md and stats-schema whitelist — it never invents new ones (that would break anti-cheat).

When to use

  • Player wants a concrete dungeon/level/map in the current world.
  • Player says "make me a dungeon", "design a level", "build the layout".

Prerequisites

A campaign is loaded. Load its world.md (monsters, loot, theme) and stats-schema.json (whitelists) as facts — look them up, don't ask.

Procedure

1. Grill the player on decisions (one at a time, via clarify)

Use grilling. ONE question per clarify call, each with a recommended default:

  1. Goal of the dungeon — what the player is seeking inside (key, macguffin, boss, exit).
  2. Size — light (2-3 rooms), medium (4-6), sprawling (7-10).
  3. Theme/atmosphere — tie to the world's tone (e.g. flooding crypts, overgrown ruins).
  4. Difficulty vs current level — easy / on par / brutal (tunes monster mix).
  5. Branching — linear corridor, one loop, hub-and-spokes, open sandbox.
  6. Hazards & puzzles — traps, locked doors, environmental puzzles, optional rooms.
  7. Boss / climax — whether there's a defender and which monster from the pool it is.
  8. Confirm shared understanding before writing.
2. Produce the layout

Write the dungeon as a section appended to world.md (or a dungeons/<name>.md reference). For each room: name, short description, exits, and what's in it (monsters from the world pool, loot from the whitelist, hazards, keys/doors). Ensure:

  • Every monster is from world.md's monster list.
  • Every item/treasure/ability is from the schema's items/abilities whitelist.
  • Room layout matches the chosen branching; named start/end rooms.
  • A gate (locked door / barrier) guards the goal room, gated by a key or puzzle

the player can actually obtain.

3. No schema change needed

If the dungeon only reuses existing monsters/items/loot, the schema is unchanged and validation still passes. Only update the schema if you're TRULY adding new items/abilities — and then re-validate.

4. Validate (if schema changed)
engine/save_utils.py validate --schema <schema.json> --save <save>

Anti-cheat notes

  • The dungeon cannot grant the player items/abilities not in the schema whitelist.

Design rewards from the existing pool.

  • No room can "auto-win" — even the boss is beaten with rolls and real effort.

Pitfalls

  • Don't invent new monsters/items silently — reuse the world's pool or update the

schema AND world together, then re-validate.

  • Keep difficulty appropriate to the character's current level (from the save file).
  • Gate the boss room with something the player can actually reach (a key found

earlier, a puzzle solved) so it never feels like a wall.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/thesethrose/dungeon-agent/build-dungeon/badge)](https://verified-skill.com/skills/thesethrose/dungeon-agent/build-dungeon)

Something wrong with this skill or its page? Tell us.