📃 Flock (Flexible Low-code Orchestrating Collaborative-agent Kits)

简体中文 | English | 日本語 | Getting Started

Tip

🎉 What's New 2024/11/12

• Intent Recognition Node: New Intent Recognition node that can automatically identify user input intent based on preset categories, supporting multi-classification routing!

• CrewAI Node Support: Now you can leverage CrewAI's powerful multi-agent capabilities in your workflows! Create sophisticated agent teams and orchestrate complex collaborative tasks with ease.

A chatbot, RAG, agent, and multi-agent application project based on LangChain, LangGraph, and other frameworks, open-source, and capable of offline deployment.

202410070930.mp4

🤖️ Overview

Work Flow

Node Types and Functions

Flock's workflow system consists of various node types, each serving a specific purpose:

1. Input Node: Processes initial input and converts it into a format the workflow can handle.
2. LLM Node: Utilizes large language models for text generation and processing.
3. Retrieval Node: Fetches relevant information from knowledge bases.
4. Tool Node: Executes specific tasks or operations, extending workflow functionality.
5. Retrieval Tool Node: Combines retrieval capabilities with tool functionality.
6. Answer Node: Generates final answers or outputs, integrating results from previous nodes.
7. Subgraph Node: Encapsulates a complete sub-workflow, allowing for modular design.
8. Start and End Nodes: Mark the beginning and end of the workflow.

Future planned nodes include:

• Intent Recognition Node
• Conditional Branch Node (If-Else)
• File Upload Node
• Code Execution Node
• Parameter Extraction Node

These nodes can be combined to create powerful and flexible workflows suitable for various complex business needs and application scenarios.

AGent Chat

Image

Knowledge Retrieval

Human in the loop (human approval or let the LLM rethink or ask human for help)

Flock aims to be an open-source platform for developing large language model (LLM) applications. It is an LLM-based application utilizing the concepts of LangChain and LangGraph. The goal is to create a suite of LLMOps solutions that supports chatbots, RAG applications, agents, and multi-agent systems, with the capability for offline operation.

Inspired by the StreetLamb project and its tribe project , Flock adopts much of the approach and code. Building on this foundation, it introduces some new features and directions of its own.

Some of the layout in this project references Lobe-chat, Dify, and fastgpt. They are all excellent open-source projects, thanks🙇‍.

👨‍💻 Development

Project tech stack: LangChain + LangGraph + React + Next.js + Chakra UI + PostgreSQL

Note

🤖 Model System

Flock supports various model providers and makes it easy to add new ones. Check out our Models Guide to learn about supported models and how to add support for new providers.

🛠️ Tools System

Flock comes with various built-in tools and supports easy integration of custom tools. Check out our Tools Guide to learn about available tools and how to add your own.

💡RoadMap

1 APP

• ChatBot
• SimpleRAG
• Hierarchical Agent
• Sequential Agent
• Work-Flow
• CrewAI Integration ---On Progress
• More muti-agent

2 Model

• OpenAI
• ZhipuAI
• Siliconflow
• Ollama
• Qwen
• Xinference

3 Ohters

• Tools Calling
• I18n
• Langchain Templates

🏘️Highlights

• Persistent conversations: Save and maintain chat histories, allowing you to continue conversations.
• Observability: Monitor and track your agents’ performance and outputs in real-time using LangSmith to ensure they operate efficiently.
• Tool Calling: Enable your agents to utilize external tools and APIs.
• Retrieval Augmented Generation: Enable your agents to reason with your internal knowledge base.
• Human-In-The-Loop: Enable human approval before tool calling.
• Open Source Models: Use open-source LLM models such as llama, Qwen and Glm.
• Multi-Tenancy: Manage and support multiple users and teams.

How to get started

1. Preparation

1.1 Clone the Code

git clone https://github.com/Onelevenvy/flock.git

1.2 Copy Environment Configuration File

cp .env.example .env

1.3 Generate Secret Keys

Some environment variables in the .env file have a default value of changethis. You have to change them with a secret key, to generate secret keys you can run the following command:

python -c "import secrets; print(secrets.token_urlsafe(32))"

Copy the content and use that as password / secret key. And run that again to generate another secure key.

1.3 Insatll postgres，qdrant,redis

cd docker
docker compose  --env-file ../.env up -d

2.Run Backend

2.1 Installation of the basic environment

Server startup requires Python 3.10.x. It is recommended to use pyenv for quick installation of the Python environment.

To install additional Python versions, use pyenv install.

pyenv install 3.10

To switch to the "3.10" Python environment, use the following command:

pyenv global 3.10

Follow these steps : Navigate to the "backen" directory:

cd backend

activate the environment.

poetry env use 3.10
poetry install

2.2 initiral data

# Let the DB start
python /app/app/backend_pre_start.py

# Run migrations
alembic upgrade head

# Create initial data in DB
python /app/app/initial_data.py

2.3 run unicorn

 uvicorn app.main:app --reload --log-level debug

2.4 run celery (Not necessary, unless you want to use the rag function)

poetry run celery -A app.core.celery_app.celery_app worker --loglevel=debug

3.Run Frontend

3.1 Enter the web directory and install the dependencies

cd web
pnpm install

3.2 Start the web service

cd web
pnpm dev

# or pnpm build then pnpm start

Star History