Tiny Chat

Installation

Tested with Python 3.10 or later

Development Installation

pip install -r requirements.txt

Package Installation

# Build the package
pip install build
python -m build

# Install the built package
pip install dist/*.whl

Web Interface Usage

Running from source (development)

streamlit run tiny_chat/main.py --server.address=127.0.0.1

only database (development)

streamlit run tiny_chat/main.py --server.address=127.0.0.1 -- --database

Running installed package

tiny-chat

only database

tiny-chat --database

MCP Usage

Claude Desktop example.

{
  "mcpServers": {
    "tiny-chat": {
      "command": "/path/to/tiny_chat/.venv/bin/tiny-chat-mcp",
      "env": {
        "DB_CONFIG": "/path/to/tiny_chat/database_config.json"
      }
    }
  }
}

OpenAI Chat API RAG Server Usage

tiny-chat-api

model: target search qdrant collection name (model change in conversation).

curl http://localhost:8080/v1/chat/completions   -H "Content-Type: application/json"   -d '{"model": "qdrant-collection-name", "messages": [{"role": "user", "content": "カレーライスの材料は？"}]}'