A comprehensive set of MCPs, specialized agents, and prompt templates for productive development with Google Antigravity.
This library is inspired by Edmund's Claude Code setup but adapted for Antigravity's agent-first architecture.
This repository provides:
- 11 Specialized Agents: Pre-configured agents for architecture, code quality, planning, and documentation
- System Prompts: Base system prompts and project brief templates
- Command Patterns: Reusable workflows for common development tasks
- MCP Configurations: Setup guides for integrating MCPs with your agents
- Clone this repo into your project or reference it when setting up a new Antigravity project
- Copy the
.antigravity/directory structure into your project - When starting a new Antigravity project, reference the appropriate agent prompts and system prompts from this library
- Customize agents and prompts as needed for your specific tech stack and project requirements
.antigravity/
├── agents/ # Specialized agent definitions
│ ├── architecture/ # System design & architecture agents
│ ├── code-quality/ # Code review, performance, security agents
│ ├── planning/ # Requirements, planning, research agents
│ └── documentation/ # Documentation & learning agents
├── commands/ # Common command/workflow patterns
├── system-prompts/ # Base prompts and templates
└── README.md # This file
- system-architect: Scalable system architecture design with trade-offs
- backend-architect: Backend systems with data integrity & security
- frontend-architect: Performant, accessible UI architecture
- requirements-analyst: Transform ideas into concrete specifications
- tech-stack-researcher: Technology choice recommendations with evidence
- refactoring-expert: Systematic refactoring and clean code principles
- performance-engineer: Measurement-driven optimization strategies
- security-engineer: Vulnerability identification and security standards
- technical-writer: Clear, comprehensive documentation generation
- learning-guide: Teaching programming concepts progressively
- deep-research-agent: Comprehensive research with adaptive strategies
- Start a new project in Antigravity
- Copy the
system-architectagent prompt into your project brief - Describe your project requirements
- Let the agent plan the architecture, then use specialized agents (backend-architect, frontend-architect) for deeper dives
- Use the
requirements-analystagent to transform your feature idea into specifications - Use
feature-plancommand to break it into implementation steps - Use
system-architector domain-specific architects to design the solution - Delegate implementation to Antigravity agents
- Use the
refactoring-expertagent to identify improvement areas - Use
performance-engineerto measure and optimize - Use
security-engineerto audit for vulnerabilities
Each project should start with the base-system-prompt and be tailored with your specific project context using project-brief-template.
This library works with:
- Filesystem MCP: For reading/writing project files
- Web Search MCP: For research and tech stack decisions
- Custom Tool MCPs: As needed for your specific project
Feel free to:
- Fork and customize agents for your tech stack
- Add new agents for domain-specific work
- Modify prompts based on your team's standards
- Extend with additional MCPs
This library is designed to work with any modern tech stack. Examples include:
- Frontend: React, Vue, Angular, Svelte
- Backend: Node.js, Python, Go, Rust, Java
- Databases: PostgreSQL, MongoDB, Firebase
- Full-stack: Next.js, SvelteKit, Remix, etc.
MIT - Use freely in your projects
Have improvements? Submit a PR or open an issue.
Created as a community resource for productive Antigravity development.