Installation¶

Requirements¶

Python 3.10+
4GB+ RAM (8GB+ for local models)

Install¶

pip install spade_llm

Verify Installation¶

import spade_llm
from spade_llm import LLMAgent, LLMProvider

print(f"SPADE_LLM version: {spade_llm.__version__}")

XMPP Server Setup¶

Built-in SPADE Server (Recommended)¶

SPADE 4.0+ includes a built-in XMPP server - no external server setup required!

# Start SPADE's built-in server (simplest setup)
spade run

The built-in server provides everything you need to run SPADE-LLM agents locally. Simply start the server in one terminal and run your agents in another.

Advanced XMPP Server Configuration¶

For custom setups, you can specify different ports and hosts:

# Custom host and ports
spade run --host localhost --client_port 6222 --server_port 6269

# Use IP address instead of localhost
spade run --host 127.0.0.1 --client_port 6222 --server_port 6269

# Custom ports if defaults are in use
spade run --client_port 6223 --server_port 6270

Alternative: External XMPP Servers¶

For production or if you prefer external servers:

Public servers: jabber.at, jabber.org (require manual account creation)
Local Prosody: Install Prosody for local hosting
Other servers: Any XMPP-compliant server

LLM Provider Setup¶

Choose one provider:

OpenAI¶

export OPENAI_API_KEY="your-api-key"

Ollama (Local)¶

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Download a model
ollama pull llama3.1:8b
ollama serve

LM Studio (Local)¶

Download LM Studio
Download a model through the GUI
Start the local server

Development Install¶

git clone https://github.com/sosanzma/spade_llm.git
cd spade_llm
pip install -e ".[dev]"

Troubleshooting¶

SPADE server not starting:

# Check if default ports are already in use
netstat -an | grep 5222

# Try different ports if needed
spade run --client_port 6222 --server_port 6269

Agent connection issues: Ensure SPADE server is running first

# Terminal 1: Start server
spade run

# Terminal 2: Run your agent
python your_agent.py

Import errors: Ensure you're in the correct Python environment

python -m pip install spade_llm

SSL errors: For development with built-in server, disable SSL verification:

agent = LLMAgent(..., verify_security=False)

Ollama connection: Check if Ollama is running:

curl http://localhost:11434/v1/models