This repository offers a production-grade starter kit for autonomous AI agent development, optimized for Google Antigravity IDE and Gemini 3 workflows. It solves the common pain point of repetitive project setup by providing a "Clone -> Rename -> Prompt" experience. Targeting engineers and power users, it delivers an enterprise-grade, transparent architecture that transforms the IDE into a knowledgeable architect, accelerating agent development.

How It Works

The project employs an "Artifact-First" protocol, generating tangible outputs like plans and logs for complex tasks. It embeds a cognitive architecture directly into project files using IDE context awareness (.cursorrules, .antigravity/rules.md), operating on a "Think-Act-Reflect" loop. This minimal, transparent architecture ("Anti-LangChain") aims to make the IDE an active participant, eliminating boilerplate configuration.

Quick Start & Requirements

• Primary install / run command: pip install -r requirements.txt or docker-compose up --build.
• Non-default prerequisites: Compatible IDE (Cursor/Google Antigravity) is key to the "zero-config" workflow. Standard Python and Docker environments are assumed.
• Estimated setup time or resource footprint: "Zero-Config" workflow implies minimal setup time.
• Links: No external quick-start or docs links provided beyond the repository itself.

Highlighted Details

• Infinite Memory Engine: Recursive summarization overcomes context limits.
• Universal Tool Protocol: Generic ReAct pattern for easy tool registration.
• LLM Support: Optimized for Gemini 3, with call_openai_chat for OpenAI-compatible APIs (Ollama, Azure).
• Zero-Config Loading: Auto-discovers tools from src/tools/ and context from .context/.
• Multi-Agent Swarm: Router-Worker pattern coordinates specialist agents.
• MCP Integration: Model Context Protocol enables seamless integration with external tools via stdio, http, sse transports.

Maintenance & Community

Key contributors include @devalexanderdaza and @Subham-KRLX. The project seeks community input for Phase 9 (Enterprise Core: Sandboxing, Orchestrated Flows) via the Issues page.

Licensing & Compatibility

The specific open-source license is not explicitly stated. Compatibility targets Google Antigravity/Cursor IDEs, supports OpenAI-compatible LLMs, and integrates broadly via MCP.

Limitations & Caveats

Phase 9 features (Sandbox Environment, Orchestrated Flows) are under development. The "zero-config" workflow depends heavily on specific IDE integrations and configurations.