From web-game-dev
Main orchestrator for web game development. Use when building, continuing, or improving a web game. Triggers on: 'build a game', 'make a game', 'game development', 'continue my game', 'improve my game'. This skill is the entry point — invoke it first.
How this skill is triggered — by the user, by Claude, or both
Slash command
/web-game-dev:game-orchestraThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
The main orchestrator workflow for web game development. This is a reference
The main orchestrator workflow for web game development. This is a reference document — the game-dev-agent reads and follows this decision tree. The agent performs the actions; this skill documents the workflow.
This skill is the entry point for all game development work. When invoked, the game-dev-agent reads this document and follows the decision tree below.
Before routing, the game-dev-agent MUST:
lessons-learned.md if it exists in the projectprogress.md if it exists in the projectNo design doc exists → invoke game-designer skill
Design approved, no code → invoke game-developer skill
Code exists, untested → invoke game-qa skill
Tests pass, needs polish → invoke game-artist skill
Code complete, pre-release → invoke game-reviewer agent
Any checkpoint reached → invoke game-psychologist agent
Cycle complete / bug fixed → invoke game-retrospective agent
Session ending → invoke game-retrospective agent
Psychologist FAIL → invoke game-retrospective agent
The game-psychologist agent is invoked at these validation gates:
Use dispatching-parallel-agents (existing CC skill) for independent tasks:
Before ANY parallel dispatch, the dispatching agent MUST output:
=== INDEPENDENCE CHECK ===
PARALLEL AGENTS: [agent-name-1], [agent-name-2]
AGENT-1 TOUCHES: [list of files/systems agent-1 will modify]
AGENT-2 TOUCHES: [list of files/systems agent-2 will modify]
FILE OVERLAP: [none | list of conflicts]
DATA DEPENDENCY: [none | "agent-2 needs agent-1 output" — describe]
DECISION: [PROCEED | SEQUENTIAL — reason]
=== CHECK COMPLETE ===
none, dispatch MUST be sequentialnone, dispatch MUST be sequential (unless agents can work from shared context — state why)Only the game-dev-agent writes to progress.md. Parallel agents report
results back to the game-dev-agent for recording.
See templates/ for progress.md format.
All dispatched agents MUST follow:
references/information-capture.md)references/validation-protocol.md)npx claudepluginhub hexcelerate/web-game-dev --plugin web-game-devProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.