1. Project Overview

Refly.AI⭐️ Open-source Vibe Workflow Platform ⭐️

Refly is an innovative open-source Vibe Workflow platform aimed at solving your most complex challenges. With just a few words, you can easily design, build, and reuse intelligent workflows, leveraging AI intelligence and infinitely extensible MCP tools to deliver high-quality results, thereby significantly improving productivity.

2. Open Source License

Licensed under Apache-2.0

3. Screenshots

4. Feature Overview

Core Features

1. Multi-threaded Dialogue System

Based on the innovative multi-threaded dialogue architecture, it supports parallel management of multiple independent conversation contexts. Through efficient state management and context switching mechanisms, it achieves complex Agentic Workflows, breaking through the limitations of traditional dialogue models.

2. Multi-model Integration Framework

• Integrates 13+ mainstream large language models, including DeepSeek R1, Claude 3.5 Sonnet, Google Gemini 2.0, OpenAI O3-mini, etc.
• Supports model mixed scheduling and parallel processing
• Flexible model switching mechanism and unified dialogue interface
• Multi-model knowledge base collaboration

3. Multi-modal Processing Capability

• File format support: PDF, DOCX, RTF, TXT, MD, HTML, EPUB, etc. 7+ formats
• Image processing: supports PNG, JPG, JPEG, BMP, GIF, SVG, WEBP, etc. mainstream formats
• Smart batch processing: supports canvas multi-element batch selection and AI analysis

4⚡️ AI-driven Skill System

Integrates Perplexity AI, Stanford Storm, etc. advanced capabilities, providing:

• Smart global search and information aggregation
• Knowledge retrieval based on vector database
• Smart question rewriting and recommendation
• AI-assisted document generation workflow

5. Context Management System

• Accurate Temporary Knowledge Base Construction
• Flexible Node Selection Mechanism
• Multi-dimensional Context Association
• Smart Context Understanding like Cursor

6 Knowledge Base Engines

• Support for Multi-source Heterogeneous Data Import
• Semantic Search Architecture based on RAG
• Intelligent Knowledge Graph Construction
• Personalized Knowledge Space Management

7✂️ Intelligent Content Collection

• Support for One-Click Content Collection from Mainstream Platforms (Github, Medium, Wikipedia, Arxiv, etc.)
• Intelligent Content Analysis and Structuring
• Automatic Knowledge Classification and Labeling
• Deep Knowledge Base Integration

8 Reference System

• Flexible Multi-source Content Reference
• Smart Context Association
• One-Click Reference Generation
• Reference Traceability Support

9✍️ AI Enhanced Editor

• Real-time Markdown Rendering
• AI Assisted Content Optimization
• Intelligent Content Analysis
• Editorial Experience like Notion

10 Code Generation Engine

• Generate HTML, SVG, Mermaid Charts, and React Applications
• Intelligent Code Structure Optimization
• Component-based Architecture Support
• Real-time Code Preview and Debugging

11 Website Visualization Engine

• Interactive Web Page Rendering and Preview
• Support for Complex Concept Visualization
• Dynamic SVG and Chart Generation
• Responsive Design Templates
• Real-time Website Prototype Design
• Modern Web Framework Integration

️ Product Roadmap

Continuously Improving Refly, Adding Exciting New Features. View the Detailed Roadmap in the Project's Complete Roadmap Documentation.

• Advanced Image, Audio, and Video Generation Capabilities
• Cross-modal Content Conversion Tool
• High-performance Desktop Client with Better Resource Management
• Enhanced Offline Functionality
• Advanced Knowledge Organization and Visualization Tools
• Collaborative Knowledge Base Functionality
• Open Standard for Third-party Plugin Development based on MCP
• Plugin Market and Developer SDK
• Least Supervised Autonomous Task Agent
• Multi-agent Collaboration System
• Complex AI Workflow Builder and API Support for Visualizing Workflows
• Advanced Integration Capabilities with External Systems
• Enhanced Security and Compliance Tools
• Advanced Team Management and Analysis

Applicable Scenarios and User Groups

Whether You're a Student Writing a Thesis, a Part-time Worker, or a Creator of Self-media, You Can Find a Suitable Scenario.

Source: Refly Official Website Documentation

5. Tech Stack

Quick Start

Before Installing ReflyAI, Please Ensure Your Machine Meets the Following Minimum System Requirements:

CPU >= 2 Cores

Memory >= 4GB

Deploy Using Docker

Deploy Your Own Feature-Rich, Unrestricted Version of ReflyAI Using Docker. The Team is Working on Updating to the Latest Version.

Start Deployment:

cd deploy/docker
cp ../../apps/api/.env.example .env # Ensure All Necessary Environment Variables Have Been Set
docker compose up -d

Deploy Using Kubernetes

Meanwhile, it also supports deploying ReflyAI in K8s clusters.

Start deployment

cd deploy/kubernetes
kubectl apply -f refly-deployment.yaml

Access http://localhost:5700 (access to Kubernetes deployment http://${HOST_IP}:30001) to start using ReflyAI.

See core deployment tutorials, environment variable configurations, and common issues in deployment tutorials.

Local development

See CONTRIBUTING for more information.

Upstream projects

Thanks to the following open-source projects that make ReflyAI possible:

1. LangChain - A library for building AI applications.
2. ReactFlow - A library for building visual workflow.
3. Tiptap - A library for building collaborative editors.
4. Ant Design - A library for building UI.
5. yjs - Provides basic support for CRDTs with state management and data synchronization.
6. React - A library for Web and native user interfaces.
7. NestJS - A library for building Node.js servers.
8. Zustand - Original and flexible state management for React.
9. Vite - Next-generation frontend tool.
10. TailwindCSS - CSS library for writing beautiful styles.
11. Tanstack Query - Library for handling frontend requests.
12. Radix-UI - Accessible React UI library for building.
13. Elasticsearch - Library for building search functionality.
14. QDrant - Library for building vector search functionality.
15. Resend - Library for building email sending functionality.
16. Other upstream dependencies.

Thank you to the community for providing such powerful and simple libraries, allowing the team to focus more on product logic implementation. Hopefully, this project will also provide a more user-friendly Vibe Workflow platform in the future.