https://github.com/neozhu/botchat
A clean, fast chat dashboard with expert personas, streaming replies, attachments, and readable code blocks.
https://github.com/neozhu/botchat
botchat expertsystem gpts
Last synced: 2 months ago
JSON representation
A clean, fast chat dashboard with expert personas, streaming replies, attachments, and readable code blocks.
- Host: GitHub
- URL: https://github.com/neozhu/botchat
- Owner: neozhu
- Created: 2026-01-19T03:00:29.000Z (5 months ago)
- Default Branch: master
- Last Pushed: 2026-03-31T10:49:24.000Z (3 months ago)
- Last Synced: 2026-04-11T03:44:05.902Z (2 months ago)
- Topics: botchat, expertsystem, gpts
- Language: TypeScript
- Homepage: https://botchat.blazorserver.com
- Size: 1.99 MB
- Stars: 9
- Watchers: 0
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Agents: AGENTS.md
Awesome Lists containing this project
README
# Botchat
**A clean, fast chat dashboard with expert personas, authenticated personal workspaces, streaming replies, attachments, and readable code blocks.**
[](https://github.com/neozhu/botchat/actions/workflows/ci.yml)
Built with **Next.js (App Router)** + **Vercel AI SDK** + **Supabase**.
## What you get
- **Expert personas**: switch between different “experts” with their own system prompts.
- **Supabase Auth**: sign up, sign in, sign out, and change password flows are built in.
- **Multi-session chat**: sessions list + titles + last-message preview.
- **User-isolated chat history**: sessions and messages are scoped to the authenticated user.
- **Streaming responses**: responsive UI while the model streams tokens.
- **Attachments**: upload images/files to Supabase Storage and send them with messages.
- **Markdown + code blocks**: readable rendering with syntax highlighting.
## Quickstart (local)
### 1) Configure env
Create `.env.local`:
```bash
cp .env.local.example .env.local
```
Required:
- `OPENAI_API_KEY` – your OpenAI API key
- `OPENAI_MODEL` – e.g. `gpt-5-mini` (model is not hardcoded in code)
Supabase (recommended for sessions/experts/attachments):
- `PUBLIC_SUPABASE_URL`
- `PUBLIC_SUPABASE_PUBLISHABLE_DEFAULT_KEY`
### 2) Run
```bash
bun install
bun dev
```
Open `http://localhost:3000`.
Authentication UI is available at `http://localhost:3000/auth`.
## Supabase setup
1. Create a Supabase project.
2. Run `supabase/schema.sql` in the Supabase SQL Editor.
3. Enable Supabase Auth email/password sign-in for the project.
4. If email confirmation is enabled, new users must activate their account from email before signing in.
5. Create a public Storage bucket named `chat-attachments`.
The database schema now assumes authenticated access:
- `chat_sessions.user_id` owns each chat session.
- `chat_messages` inherit ownership through `session_id`.
- Row Level Security allows users to read/write only their own sessions and messages.
- `experts` remain global/shared and are not user-owned.
## Docker
This repo includes a production Docker build for the Next.js app:
```bash
docker compose up -d --build
```
App is exposed at `http://localhost:3202` (mapped from container `3000`).
Tip: for Docker envs, set `OPENAI_API_KEY`, `OPENAI_MODEL`, `PUBLIC_SUPABASE_URL`, and `PUBLIC_SUPABASE_PUBLISHABLE_DEFAULT_KEY` in your deployment platform or `docker-compose.yml`.
## Project structure
- `app/` – Next.js routes (UI + API routes)
- `components/botchat/` – chat UI (sessions, experts, input, panels)
- `components/ai-elements/` – AI-friendly UI primitives (rendering, message building blocks)
- `lib/` – shared utilities + Supabase clients
- `supabase/schema.sql` – database schema + RLS policies
## Common scripts
```bash
bun dev # local dev
bun run build # production build
bun start # run production server on :3000
```