team-level

Orchestrates a multi-agent team to produce complete game level design documents.

Updated Aug 2, 2026
One-click install
npx skills add https://github.com/claretagrapelike32/codex-ai-game-studio --skill team-level-claretagrapelike32
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: team-level
Source: https://github.com/claretagrapelike32/codex-ai-game-studio/tree/main/plugins/ai-game-studio/skills/team-level
Command: npx skills add https://github.com/claretagrapelike32/codex-ai-game-studio --skill team-level-claretagrapelike32

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Designing a game level requires input from narrative, world-building, art, systems, accessibility, and QA disciplines, and coordinating all of them manually is slow and error-prone. This Skill orchestrates specialized subagents in a gated sequence to produce a complete, reviewed level design document. ## Core Features & Use Cases - Multi-Agent Orchestration: Spawns narrative-director, world-builder, level-designer, systems-designer, art-director, accessibility-specialist, and qa-tester subagents in a five-step pipeline with user approval gates between steps. - Review Modes: Supports full, lean, and solo review modes via a --review argument or production/review-mode.txt to control how many director gates run. - Dependency and Accessibility Checks: Detects unresolved adjacent-area references and surfaces BLOCKING accessibility concerns before finalizing the design. - Use Case: A designer runs the skill with the argument forest dungeon; the skill reads the GDD and pillars, coordinates all subagents, and writes an approved level design doc to design/levels/forest-dungeon.md. ## Quick Start Ask the AI to run the team-level skill for a target area such as 'tutorial' or 'final boss arena' and approve each gated step as the subagent proposals are presented.

Frequently Asked Questions about team-level

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I design a game level with multiple AI agents?

Invoke the skill with the target area name as an argument, such as 'forest dungeon'. It spawns narrative, world-building, level design, systems, art, accessibility, and QA subagents in sequence, pausing at gates for your approval before writing the final document.

What review modes does the level design workflow support?

Three modes are supported: full runs all director gates, lean skips director gates except phase-gate types, and solo skips all agent gates. Set the mode with a --review argument or a production/review-mode.txt file; the default is lean.

Where is the final level design document saved?

The compiled document is written to design/levels/[level-name].md by the level-designer subagent after you approve the write request. The orchestrator itself never writes files directly.

What happens if a referenced adjacent level does not exist?

The skill flags the missing area and asks whether to proceed with an UNRESOLVED placeholder reference or pause to design that area first. It never invents content for the missing adjacent area.

How are accessibility issues handled during level design?

An accessibility-specialist subagent reviews navigation clarity, colorblind safety, and cognitive load, rating concerns as BLOCKING, RECOMMENDED, or NICE TO HAVE. BLOCKING issues must be acknowledged before the workflow proceeds to QA planning.

What happens when a subagent fails or returns BLOCKED?

The skill surfaces the blocker immediately, checks whether later phases depend on it, and offers options to skip, retry with narrower scope, or stop. A partial report with all completed work is always produced.