stableagents-ai 0.1.4

A framework for stable AI agents

StableAgents AI

A framework for building the Linux kernel of AI agents - providing the core infrastructure and system-level capabilities that enable reliable, secure, and efficient AI agent operations.

Installation

pip install stableagents-ai

Or with Poetry:

poetry add stableagents-ai

For local LLM support:

pip install stableagents-ai[local]
# or with Poetry
poetry add stableagents-ai -E local

Quick Start

# Using the Python API
from stableagents import StableAgents

agent = StableAgents()
agent.set_api_key('openai', 'your-api-key')
agent.set_active_ai_provider('openai')

response = agent.generate_text("Tell me about AI agents")
print(response)

# Using with a local model
agent = StableAgents()
agent.set_local_model()  # Uses default model location
# or specify a model path
# agent.set_local_model("/path/to/your/model.gguf")
response = agent.generate_text("Tell me about AI agents")
print(response)

Command Line Interface

StableAgents comes with a simple CLI that you can run from anywhere:

# Run the CLI directly with any of these commands
stableagents
stableagents-ai
run-stableagents

# Run with a specific model and API key
stableagents --model openai --key your-api-key

# Run with a local model
stableagents --local

# Run with a local model from a specific path
stableagents --local --model-path /path/to/your/model.gguf

Once in the CLI, you can:

• Chat with the AI directly by typing any text
• Use commands like memory, control, and provider
• Type help to see all available commands

Features

• Multiple AI provider support (OpenAI, Anthropic, etc.)
• Local model support for offline usage (via llama-cpp-python)
• Memory management
• Computer control capabilities
• Simple but powerful CLI
• Logging system

Local Models

StableAgents supports running LLM inference locally using llama-cpp-python. To use local models:

1. Install the local dependencies: pip install stableagents-ai[local]
2. Download a compatible GGUF model file (e.g., from TheBloke on Hugging Face)
3. Place the model file in ~/.stableagents/models/default/ or specify the path when loading

# Load a model from a specific path
agent.set_local_model("/path/to/your/model.gguf")

# Or run the CLI with local mode
# stableagents --local

# Or specify a custom model path in the CLI
# stableagents --local --model-path /path/to/your/model.gguf

The framework will automatically search for a .gguf file in the default directory if no path is specified.

License

MIT