answeryt/Jumping-Agent-platform
Build your own AI agent through gameplay.
What it solves
It lowers the barrier to entry for beginners to create AI agents by replacing traditional, complex 2D workflow diagrams (with lines and arrows) with a gamified, spatial 3D interface. It transforms abstract agent logic into a tangible process where users can visually build and track the agent's progress through a "jumping" interaction.
How it works
The platform uses a Three.js-based frontend to present agent construction as a game-like experience. Users configure nodes and workflows through this spatial interface, which then communicates with an orchestrator backend. This backend utilizes an agent_builder for structural templates and a back_agent (a ReAct-based agent) to handle code completion and rewriting based on user requirements. Once built, the agent is deployed in a dedicated workspace and can be connected to WeChat via a specialized bridge connector for real-time interaction.
Who it’s for
Non-experts and beginners who want to build their own AI agents without needing to understand complex workflow orchestration or write extensive code.
Highlights
- Gamified UI: Uses a "jump-style" 3D interaction to visualize agent workflows.
- WeChat Integration: Supports binding agents to WeChat accounts via QR code for direct messaging.
- Automated Code Generation: Employs a ReAct agent to automatically complete and refine the agent's underlying code based on user input.
- Template-Based Building: Provides various workflow templates (sequential, routing, parallel, etc.) to accelerate development.
Related
- Project
- Project
- Project
- Project