Skip to main content

AI-driven discussions and multi-agent debates

Project description

superchat

AI-driven discussions and multi-agent debates. A terminal-based CLI application for communicating with multiple AI models through OpenRouter's API.

Features

  • Interactive chat with any open source AI model on OpenRouter
  • Add multiple agents to a chat to start a debate between agents

Requirements

  • Python 3.8+ - Check with python3 --version
  • pip - Usually comes with Python, check with pip --version
  • OpenRouter API key - Get one at https://openrouter.ai/keys

Installing missing dependencies on macOS:

  • Python: brew install python or download from python.org

Installation

Install superchat with one command

pip install superchat

Test that it worked

superchat --help

If you see a dialog with instructions, then you're ready to use superchat

If superchat command isn't found:

Run this command to add the install location to your PATH:

echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc && source ~/.zshrc

Configuration

You'll need an OpenRouter API key to use superchat. The program will guide you through this setup automatically on first run.

Getting your API key:

  1. Go to https://openrouter.ai/keys
  2. Sign up for a free account
  3. Create a new API key (starts with sk-or-)
  4. Add credits to your account at https://openrouter.ai/credits

Setup:

When you first run superchat, it will detect that you don't have an API key and automatically prompt you to enter it. The key will be securely saved to your ~/.env file for future use.

Manual Setup (Optional):

If you prefer to set up the API key manually:

echo "OPENROUTER_API_KEY=ADD-YOUR-KEY-HERE" > ~/.env

Usage

After installation, run from any terminal:

superchat

Or use the short alias:

sc

Debate Modes

Superchat supports two different debate modes when using multiple AI models:

Default Mode

In default mode, all AI agents join a shared conversation immediately. Each user message triggers responses from all agents in a round-robin format.

# Start default mode with multiple models
superchat -m llama claude deepseek

Staged Mode

In staged mode, you first have individual 1:1 conversations with each agent, then transition to a team debate with shared context from all conversations.

# Start staged mode with multiple models
superchat --flow staged -m llama claude deepseek

Staged Mode Commands:

  • /promote - Promote current agent to team debate and move to next agent

Staged Mode Benefits:

  • Get individual perspectives before group discussion
  • Agents share context from 1:1 conversations in team phase

Debate Rounds

Control how many times each agent responds per user message in team debates:

# Set debate rounds (1-5, default: 1)
superchat -r 3 -m llama claude

Updating

To update to the latest version:

pip install --upgrade superchat

Roadmap

Planned features:

- Web search
- Image processing
- Voice mode

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

superchat-0.10.tar.gz (48.2 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

superchat-0.10-py3-none-any.whl (54.7 kB view details)

Uploaded Python 3

File details

Details for the file superchat-0.10.tar.gz.

File metadata

  • Download URL: superchat-0.10.tar.gz
  • Upload date:
  • Size: 48.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.1

File hashes

Hashes for superchat-0.10.tar.gz
Algorithm Hash digest
SHA256 5452f0b59a58c22aa57e275d9ca36cbeb08bbbd09a4b705ea4e2d3b5f332ebca
MD5 bde41533ec12798cfe13525cfed2e263
BLAKE2b-256 f85f4c7fd6d8511672c346a2d0ab45b9f83959ec21cf9b8340c3f9bcea197a85

See more details on using hashes here.

File details

Details for the file superchat-0.10-py3-none-any.whl.

File metadata

  • Download URL: superchat-0.10-py3-none-any.whl
  • Upload date:
  • Size: 54.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.1

File hashes

Hashes for superchat-0.10-py3-none-any.whl
Algorithm Hash digest
SHA256 b07af97b16293544f4ba631cc6e86d46b92371d905c7a960977d5431f046958c
MD5 76530911f862026725f3eac7b930b37a
BLAKE2b-256 1a5219b4bdb9d8fd2dedb8ef6e642fa4a9dfb08743da7d78b5ec6886e7300cef

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page