Helios

Written by Helios

A command-line AI assistant for software engineering tasks with code generation, git integration, and repository management capabilities.

Table of Contents

• Features
• Usage
• Configuration
• Demo

Features

The AI Code Assistant provides the following core functionalities:

• Code generation and modification with context-awareness
• Git integration
• File management commands
• Repository indexing and refresh
• GitHub API interactions
• Project structure creation and validation
• AI PR reviews
• AI Repository summaries

Usage

Prerequisites: Python, Ollama

Run Helios

# Clone repo
git clone https://github.com/viditvidit/helios.git

# Run in development mode
pip install -e .

# Run this to use Helios
helios

Create a .env file

GITHUB_TOKEN = <YOUR GIT TOKEN>
GITHUB_USERNAME = <YOUR GIT USERNAME>

OLLAMA_HOST = http://localhost:11434 # or your custom host link

Change or Add your model of choice

Make changes in: models.yaml

# Change here
models: 
  gemma3: # Set custom name
    name: 'gemma3:27b-it-q4_K_M' # Enter model name as seen in ollama
    <<: *common_settings

Know all features:

/help # For exploring all features of Helios

Agent Prompt Example

/knight create a .NET console application named 'dotnet-weather-cli'. The app should take a city name as a command-line argument, make a GET request to a public weather API (like Open-Meteo), and print the current temperature for that city. Use the `HttpClient` class for the API call and `Newtonsoft.Json` to parse the response.

Configuration

Modify configuration settings in configs/models.yaml to customize:

• AI model parameters
• Agent instructions and principles
• Code generation guidelines
• Project structure templates

Demo

Landing:

File Creation

File mentions

GitHub Features:

Smart Review Open PRs

Knight Agent

Agent prompt and plan

Execution

Git Push

Other Git creations:

AI Features:

PR Review Repo Summary