JellyChat
🪼 AI chatbot for the DeFiChain ecosystem.
Install / Use
/learn @0ptim/JellyChatREADME
JellyChat is a project that aims to create a chatbot that can answer questions about DeFiChain and its ecosystem.
❗ There are many people which want to learn more about DeFiChain and its ecosystem. However, it is not always easy to find the right information. Also, many people struggle to take in information from various articles in a wiki.
✨ JellyChat aims to solve this by providing the ability for users to interactively ask questions and have an ongoing conversation.
Main Objectives
- Make learning about DeFiChain fun.
- Be able to answer general questions about DeFiChain and its ecosystem.
- Be able to answer questions involving on-chain data.
- Behave like Jelly (the DeFiChain mascot).
- Feel like a friend that you can ask anything.
Architecture
This gives a high-level overview of the architecture. For more details, please refer to the individual components.
Accounts and History
Identification of accounts is done with a user_token which the client needs to store. The client then posts this user_token to /history and gets the history back if the account exists. If the account does not exist, a new one is created automatically.
Input/Output Flow
The client posts a user message and a user_token to /user_message. A LangChain agent will be created for the specific user_token and the memory is filled from the history. The agent will then analyze the user's message and return an answer. The answer is then returned to the client and also added to the history.
Key Components
The Repository contains the following key components.
🌐 /backend
The backend is a Flask API that receives questions and returns answers. It uses a LangChain agent to analyze the question and then uses various tools to best answer the question.
💽 /data
We use Supabase to store:
- Users
- Messages
- QA data
- Embeddings
⚡️ /job
Contains a script that is executed daily at 02:45 UTC via a GitHub action. However, it only runs if there have been changes to the /docs directory within 0ptim/DeFiChainWiki in the past 24 hours. If changes are detected, the script scrapes the DeFiChainWiki to generate new embeddings. These embeddings are then stored in Supabase in the table embeddings.
Workflow and Environments
The main branch resembles the production code. When a feature is ready, a pull request is created onto main. This will trigger the staging actions so the changes can be tested on the staging environment.
- 💽 Staging Database
- Endpoint: https://iajfqvpslzrtmvekqwpv.supabase.co
- 🌐 Staging Backend
- Endpoint: https://jellychat-staging.fly.dev
When the pull request is merged (changes are pushed to main), the production action will be triggered and the changes will be deployed to production.
- 💽 Production Database
- Endpoint: https://cofmxsabxteiidryklyg.supabase.co
- 🌐 Production Backend
- Endpoint: https://jellychat.fly.dev
Related Skills
claude-opus-4-5-migration
109.4kMigrate prompts and code from Claude Sonnet 4.0, Sonnet 4.5, or Opus 4.1 to Opus 4.5
model-usage
349.0kUse CodexBar CLI local cost usage to summarize per-model usage for Codex or Claude, including the current (most recent) model or a full model breakdown. Trigger when asked for model-level usage/cost data from codexbar, or when you need a scriptable per-model summary from codexbar cost JSON.
feishu-drive
349.0k|
things-mac
349.0kManage Things 3 via the `things` CLI on macOS (add/update projects+todos via URL scheme; read/search/list from the local Things database)
