Report Abuse

Basic Information

Agentok Studio is a visual development tool for creating, visualizing and running multi-agent applications built on the AG2 agent framework. It provides a diagram-based UI to compose agent workflows, define conversation relations between nodes, configure tool calls on edges, and produce native Python code that depends only on the official ag2 library. The repository contains a Next.js frontend and a FastAPI backend, examples and project scaffolding, Docker and non-Docker run instructions, and integration points for Supabase authentication and storage. The project emphasizes visibility into the generated code, execution logs and data structures to aid debugging. The README notes experimental additions such as a DeepSeek Assistant Node and the ability to configure LLM models for conversable nodes. A separate backend effort named VibeX is referenced as the planned new backend.

Links

Categorization

App Details

Features
Drag-and-drop visual flow editor that maps AG2 concepts into nodes and edges for easy construction of agent workflows. Tool editor with variable configuration so tools can be defined and attached to conversation edges. Diagram-to-code generator that emits self-contained Python programs relying on the ag2 library. Full visibility into flow data and execution logs including stdout and stderr for diagnosis. Built frontend (Next.js) and backend (FastAPI) with Docker compose support and standalone development instructions. Theme toggle (light/dark), chat interface for testing flows, sample project templates and OpenAPI docs for the backend. Observability integration via AgentOps and notes about configuring model and LLaVa-related tokens. RAG was explicitly removed from the project.
Use Cases
Agentok Studio lowers the barrier to building complex multi-agent applications by turning AG2 constructs into an interactive visual environment that non-expert developers and teams can use to prototype and iterate. The tool generator produces native Python code so flows can be exported, run and debugged outside the UI while preserving full traceability of internal data and logs. Docker and local run instructions simplify getting a development instance up and running. The tool editor and edge-level tool configuration streamline integrating external tool calls into conversations. The repository also documents required environment variables, Supabase setup for auth and storage, and observability hooks so teams can deploy, monitor and extend agent workflows in familiar environments.

Please fill the required fields*