chewt/ai-assistant
1
0
Fork 0
Code Issues Pull requests Actions Packages Projects Releases Wiki Activity
A Python script that enables you to chat with an intelligent assistant (LLaMA) in your terminal. The script uses the ollama library to interact with the LLaMA model and provide a user-friendly interface.
43 commits 2 branches 0 tags 211 KiB
Python 100%
Go to file
Hayden Johnson 780b755837 Add SQLite database support for conversation history 
- Introduce SQLite database to store conversation history
- Implement functions to save and load conversations from the database
- Replace previous file-based history storage with database-backed solution
- Add new command to list and resume previous conversations
- Update chat logic to generate topics for saved conversations
- Add database initialization and management functions
- Modify command line interface to support resuming conversations
2025-05-17 10:18:50 -07:00
.gitignore Update gitignore 2024-09-25 11:58:55 -07:00
.python-version use refactored version by default now 2025-05-16 07:44:27 -07:00
assistant.py Add SQLite database support for conversation history 2025-05-17 10:18:50 -07:00
pyproject.toml use refactored version by default now 2025-05-16 07:44:27 -07:00
README.md use refactored version by default now 2025-05-16 07:44:27 -07:00
requirements.txt add requirements.txt 2025-03-16 12:10:15 -07:00
uv.lock use refactored version by default now 2025-05-16 07:44:27 -07:00

README.md

Assistant

A terminal-based chat interface with an intelligent assistant.

Overview

This script provides a simple way to interact with an AI assistant in your terminal. It uses the Ollama API to generate responses and can handle markdown code blocks, syntax highlighting, and shell commands.

Features

  • Terminal-based chat interface
  • Supports markdown code blocks
  • Syntax highlighting using Pygments
  • Handles shell commands
  • Can copy highlighted code blocks to clipboard
  • Allows follow-up questions with --follow-up argument
  • Customizable model and temperature with --model and --temp arguments

Usage

Command Line Arguments

The script supports the following command line arguments:

  • --follow-up: Ask a follow-up question when piping in context
  • --copy: Copy a codeblock if it appears
  • --shell: Output a shell command that does as described
  • --model: Specify model (default: llama3.1:8b-instruct-q8_0)
  • --temp: Specify temperature (default: 0.2)
  • --host: Specify host of Ollama server

Piped Input

You can pipe input to the script by running it without any arguments. The script will read from standard input and handle it as a single question.

Non-Piped Input

Run the script with no arguments to enter interactive mode. Type your questions or commands, and the assistant will respond accordingly.

Installation

You can install the dependencies in a virtual environment using uv

uv sync

You also need to set up an Ollama server on your local machine. Please refer to the Ollama documentation for instructions.

Running the Script

Run the script with Python:

python assistant.py