YggDonchitos/Claude-Code-Game-Studios
↗ GitHubTurn Claude Code into a full game dev studio — 48 AI agents, 36 workflow skills, and a complete coordination system mirroring real studio hierarchy.
7,792
Stars
1,154
Forks
88
Watchers
3
Open Issues
Safety Rating A
The repository is a well-structured collection of markdown agent definitions, bash hook scripts, and documentation templates. No secrets, malicious patterns, suspicious dependencies, or prompt injection attempts were identified. The hook scripts perform standard validation operations (JSON linting, git checks, file naming conventions) and include explicit safety rules blocking dangerous operations like force pushes and `rm -rf`. The project appears to be a legitimate, well-intentioned open-source template for game development with AI assistance.
ℹAI-assisted review, not a professional security audit.
AI Analysis
Claude Code Game Studios is a comprehensive template and configuration system that transforms a single Claude Code AI session into a structured, hierarchical game development studio. It provides 48 specialized AI subagents organized into a three-tier studio hierarchy (directors, department leads, and specialists), 37 slash-command workflow skills, 8 automated validation hooks, 11 path-scoped coding rules, and 29 document templates. The system mirrors real-world studio organization with defined delegation paths, escalation procedures, and quality gates — covering game design, programming, art direction, audio, narrative, QA, production, and engine-specific workflows for Godot 4, Unity, and Unreal Engine 5. It is explicitly collaborative rather than autonomous: agents ask questions, present options, and require user approval before acting.
Use Cases
- Structuring a solo or small-team game development project with AI-assisted workflows covering the full game dev lifecycle
- Using specialized AI agents for domain-specific tasks such as GDD writing, code review, balance checking, sprint planning, and release management
- Enforcing coding standards and documentation requirements via automated hooks and path-scoped rules in a Claude Code session
- Coordinating multi-agent AI workflows across design, programming, art, audio, and QA disciplines for a game project
- Bootstrapping game projects with professional templates for design documents, ADRs, economy models, and sprint plans
Tags
Security Findings (4)
No hardcoded secrets, API keys, or tokens were found. The repository is a shell-script and markdown template system with no credential storage.
No malicious code patterns detected. Hook scripts perform standard git validation, file naming checks, and session logging. The `validate-commit.sh` hook checks for hardcoded values in user code, which is a protective measure rather than a threat.
No package manifests (package.json, requirements.txt, etc.) with versioned dependencies are present in the described structure. The only runtime dependency is the Claude Code npm package installed externally by the user, which is not bundled.
No prompt injection attempts detected. The README and agent configuration content describe legitimate AI orchestration patterns without any embedded instructions designed to manipulate AI analysts.
Project Connections
skyclaw
→Both projects implement multi-agent AI orchestration systems with hierarchical agent coordination. Skyclaw targets general autonomous agent runtimes across multiple platforms, while Claude Code Game Studios targets game development workflows within Claude Code specifically.
clawvault
→ClawVault provides persistent markdown-native memory and session lifecycle management for AI agents, which would directly address the context continuity challenges Claude Code Game Studios faces between sessions (e.g., preserving sprint context and project state across multiple Claude Code sessions).
code-review-graph
→code-review-graph's MCP-based blast-radius analysis and codebase knowledge graph would complement the code-review and tech-debt slash commands in Claude Code Game Studios, giving the programming agents better structural context about the game codebase.
GitNexus
→GitNexus provides MCP-exposed code intelligence (symbol context, dependency graphs, semantic search) that would enhance the programming specialist agents in Claude Code Game Studios when performing code reviews, refactoring, or impact analysis on game source code.
kavach
→Kavach's EDR capabilities for monitoring and restraining AI agent file system operations would add an additional safety layer on top of Claude Code Game Studios' own permission rules, particularly useful given the system spawns 48 agents with defined file-domain boundaries.