Engineering
app-builder avatar

app-builder

Full-stack application orchestrator that analyzes natural language requests to determine tech stacks, scaffold projects, and coordinate specialized development agents.

Introduction

The App Builder is the primary orchestration engine within the Antigravity system, designed to transform abstract natural language requirements into fully functional, production-ready full-stack applications. By leveraging a comprehensive set of predefined templates and a network of specialized agents, this skill automates the entire software development lifecycle from initial discovery to final deployment.

  • Intelligent project detection using keyword matrices to identify requirements such as SaaS, REST APIs, mobile apps, or static sites.

  • Automated scaffolding based on 12+ pre-configured stacks including Next.js, Prisma, FastAPI, Flutter, Electron, and Chrome extensions.

  • Orchestration of 16 specialized agents, including project-planner, backend-specialist, frontend-specialist, and database-architect, to ensure cohesive development.

  • Context-aware file handling through strict selective reading rules, ensuring agents only access necessary documentation and project files.

  • Feature analysis and error handling integrated directly into the development workflow for iterative improvements.

  • Users should provide clear, intent-driven requests (e.g., 'Make an Instagram clone with photo sharing') to allow the builder to accurately select the optimal tech stack.

  • The system utilizes a modular skill directory; ensure the environment is correctly set up with global workflows and agent configurations before initiation.

  • Always refer to project-detection.md for supported project types and tech-stack.md for 2025 default standard technology choices.

  • Constraints include maintaining clean code standards, adherence to defined agent execution pipelines, and ensuring the local environment matches the expected Antigravity directory structure (e.g., ~/.gemini/antigravity/).

  • Inputs consist of natural language prompt strings; outputs are structured project files, API route definitions, database schemas, and initial UI component scaffolds.

Repository Stats

Stars
131
Forks
21
Open Issues
1
Language
Python
Default Branch
main
Sync Status
Idle
Last Synced
May 3, 2026, 08:06 PM
View on GitHub