Introduction

🎉 Welcome to Fluctlight! Your Gateway to Intelligent Chat Agents 🎉

The ultimate framework designed for crafting task-based chat agents. Seamlessly configure and deploy your agents across leading platforms like Discord, Slack, and the web with ease.

🚀 Why Choose Fluctlight? 🚀

• 🌐 Cross-Platform Deployment: Effortlessly reach users on Discord, Slack, or any web interface.
• 🧠 Advanced Capabilities:
  • RAG Retrieval: Leverage the power of cutting-edge retrieval-augmented generation.
  • Character AI: Create character agents with distinct personas.
  • Workflow Task Agent Support: Enhance AI agents with workflow task support, ideal for shopping assistant bots and customer service bots.
  • Dynamic Configuration: Configure chatbot agents via static YAML files or REST API, and store your agent configurations in a backend database.
  • Speech & Image Understanding: Boost your agent's capabilities with integrated speech recognition and image comprehension.

Transform your interaction experiences with Fluctlight, where innovation meets simplicity and versatility. Get started today and redefine what's possible!

Design

Details design blueprint

Spotlights

RAG(Retrieval-Augmented Generation)	Image/Audio Support	Workflow base agent

Development

Prerequisite

1. Docker is required for local testing
2. asdf-vm is prefered to install dev env.

Setup

Checkout the project with submodules included.

git clone --recurse-submodules git@github.com:fengyu05/fluctlight.git

Then run the setup env, it will run asdf plugins install to install 'python', 'poetry' and 'precommit' to a desire versions.

make install

Run the Server

To get started, you will need to set up a Slack or Discord bot by following the official tutorials, or our tutorial. Then, you can run the server locally using Docker by following these steps:

1. Rename env.template to .env.
2. Fill in the required APIKEY and other configurations as explained below.
3. Run make slack-server or make discord-server to bootstrap the server.

Features developement progress

progress

Tooling

Running the make command will list all available tasks:

❯ make
===== All tasks =====
slack-server                   start slack server
discord-server                 start discord server
debug-server                   start app server with debug mode
api-server                     start app server
test                           Test
test-int                       Integration test
docker-compose-build           Build the app
docker-compose-up              Run app with rebuild
docker-compose-bash            Connect to a bash within the docker image
docker-test                    Run unit test
docker-test-int                Run integration test
requirements.txt               Export requirements.txt from pyproject.toml
requirements-dev.txt           Export requirements-dev.txt from pyproject.toml
alembic-init                   initial alembic db env
alembic-revision               Create a new alembic revision file[within a docker app env], e.g make alembic-revision REV_LOG="add user table" DB_APP=app
alembic-upgrade                Apply alembic revision, e.g make alembic-upgrade REV=xyz DB_APP=app

Configuration

Below explain the main configuration in .env

These are API keys may be used

OPENAI_API_KEY # OpenAI API Key, required as default
FIREWORKS_API_KEY # Optional, alternative to the OPENAI Api key.
SLACK_APP_OAUTH_TOKENS_FOR_WS # Optional, for Slack-bot
SLACK_APP_LEVEL_TOKEN # Optional, for Slack-bot
DISCORD_BOT_TOKEN # Optional, for Discord-bot

These are the main funtional switch

BOT_CLIENT="DISCORD|SLACK" # Config whether to bind the Discord or Slack frontend when app start.

#For all special usecase, if you only need a general purpose chatbot, you won't be using these options
INTENT_CHAR_MATCHING=false  # Whether to enable character agent matching
INTENT_LLM_MATCHING=false  # Whether to enable multiply purpose agent matching
INTENT_EMOJI_MATCHING=false # Whether to enable emjoi to agent mapping

These are the model config

OPENAI_GPT_MODEL_ID="gpt-4o" # To config what model use for all the glue LLM call. Can use mini to save cost.
OPENAI_CHATBOT_MODEL_ID="gpt-4o" # To config what model use for the character tone.

These are for debugging

LANGCHAIN_TRACING_V2 # To enable tracing using Langchain
DEBUG_MODE # Debug mode(more verbose logs)

More spec please see env.template

Deployment

Botchan can be run locally on your development machine or laptop. For a 24/7 online deployment option:

• Consider deploying the application in a cloud Kubernetes environment, as it is already containerized.

Discord Bot Setup

Refer to the Discord Bot Setup Tutorial.

Slack Bot Setup

Follow the Slack Bot Setup Tutorial.

Security and Privacy Reminder

• Log chat messages only when the debug flag is enabled.
• The repository is not designed for production use, so exercise caution and use at your own risk.

FAQ

1. makefiles is missing

make
Makefile:5: makefiles/alembic.mk: No such file or directory

please ensure you checkout the project with --recurse-submodules because make file is make external module. or at least run make install once.