FEAGI Python SDK
Build AI agents that learn like biological brains
What is FEAGI?
FEAGI (Framework for Evolutionary Artificial General Intelligence) is a biologically inspired, modular neural execution engine designed for embodied AI and robotics. FEAGI enables perception, cognition, and control through spiking neural circuits across simulated and physical embodiments, with a strong emphasis on real-time interaction, modularity, and cross-platform deployment.
FEAGI includes a 3D brain visualizer for real-time neural activity and official open source SDKs for Python (this repository), Rust, and Java. Together they support a growing ecosystem of reusable neural components ("brains"), tools, and integrations for robotics and physical AI.
Neurorobotics Studio is FEAGI's desktop application for working with brains, simulations, and embodiments in a unified environment.
The FEAGI Python SDK
The FEAGI Python SDK provides the tools you need to:
- Connect robots and devices to FEAGI's neural network
- Build learning agents for robots, simulators, and games
- Visualize neural activity in real-time with Brain Visualizer
- Control and manage FEAGI from Python code
- Interface with diverse embodiments through standardized communication protocols
Quick Start
pip install feagi
feagi start
feagi bv start
That installs the feagi package (engine, Brain Visualizer, and SDK), starts FEAGI with default configuration, and opens the Brain Visualizer. The same flow works on Linux, macOS, and Windows.
Installation
Quick Start uses feagi. If you need a smaller install without Brain Visualizer, use feagi-core instead. Imports are the same for both (from feagi import ...).
| Command | What you get |
|---|---|
pip install feagi |
Full install: neuronal engine, Brain Visualizer (adds ~196MB of platform binaries), Rust-backed Python bindings, and the agent SDK. Recommended for most users. |
pip install feagi-core |
Engine, bindings, and agent SDK only—no Brain Visualizer. Recommended for production, CI, or headless setups. |
Key Concepts
-
Neuromorphic by Design – FEAGI is built as a neuromorphic framework inspired by biological neural computation. While it currently runs on conventional CPUs and GPUs, native support for neuromorphic hardware is a near-term roadmap item, enabling direct execution on event-driven, spike-based accelerators as they mature.
-
Embodied Intelligence First – FEAGI is designed to control bodies (robots, agents, simulations), not just process static data.
-
Spiking Neural Networks (SNNs) – Uses event-driven neuron firing rather than frame-based inference.
-
Modular Neural Architecture – Neural circuits can be composed like building blocks (Lego-like micro-circuits).
-
Real-Time Closed Loop – Continuous perception → cognition → action loop.
-
Cross-Simulator & Hardware Support – One brain, many bodies.
Documentation
Configuration Management
Initialize FEAGI environment with default configuration:
feagi init
This creates:
- Configuration:
~/.feagi/config/feagi_configuration.toml - Genomes directory:
~/Documents/FEAGI/Genomes/(macOS/Windows) or~/FEAGI/genomes/(Linux) - Connectomes directory:
~/Documents/FEAGI/Connectomes/or~/FEAGI/connectomes/ - Logs and cache directories
For complete configuration options and customization, see DEPLOY.md.
Start FEAGI Engine from Python
from feagi.engine import FeagiEngine
engine = FeagiEngine()
engine.load_config() # Uses default config
engine.load_genome("my_brain.genome") # Loads from genomes directory
engine.start()
To restore a trained connectome instead:
engine = FeagiEngine()
engine.load_config()
engine.load_connectome("trained_brain.connectome")
engine.start()
Genome and connectome startup artifacts are mutually exclusive. The equivalent command-line forms are:
feagi start --config ~/.feagi/config/feagi_configuration.toml --genome my_brain.genome
feagi start --config ~/.feagi/config/feagi_configuration.toml --connectome trained_brain.connectome
Download the active brain using the API connection from the selected config. The timeout is explicit so large artifacts do not depend on an implicit network limit:
feagi download genome --output current.genome --timeout <SECONDS>
feagi download connectome --mode full --output current.connectome --timeout <SECONDS>
feagi download connectome --mode lite --output current-lite.connectome --timeout <SECONDS>
Genome artifacts contain JSON but must use .genome. The SDK rejects .json
genome filenames; connectome artifacts continue to use .connectome.
SDK Architecture
feagi/
├── agent/ # Agent framework (BaseAgent)
├── pns/ # Peripheral Nervous System (communication)
├── engine/ # Engine control
├── config/ # Configuration management
├── paths/ # Cross-platform path utilities
├── cli/ # Command-line tools
├── genome/ # Genome artifact I/O, validation, and live operations
├── connectome/ # Connectome artifact I/O, migration, and live operations
└── packaging/ # Marketplace packages (coming soon)
Examples
See examples/ for complete agent implementations:
- Basic sensory agent
- Robot agent (SDK-based)
- Simulator agent (Webots)
- Vision processing
Community & Support
- Discord: Join our community
- Issues: Report bugs
- Neurorobotics Studio: Design and deployment desktop application
- Homepage: feagi.org
Requirements
- Python 3.10 or higher
- Works on Linux, macOS, and Windows
License
Apache 2.0 - See LICENSE for details.
Copyright 2016-2025 Neuraville Inc. All Rights Reserved.
About Neuraville
FEAGI is developed by Neuraville, a company focused on democratizing robotics and enabling the next generation of embodied AI through modular, biologically inspired intelligence systems.
1 maintainer
Metadata
1 maintainer
Release files for feagi-core 2.1.45
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| feagi_core-2.1.45.tar.gz | 29.3 MB | Without provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| feagi_core-2.1.45-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 58.8 MB
Release files / feagi_core-2.1.45.tar.gz
| Download URL | feagi_core-2.1.45.tar.gz |
|---|---|
| Size | 29.3 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
e67c30b6b7fed39a3a3f45192816ed248c5e5d95190d6d9d60c166faec2f1446
|
|
BLAKE2b-256 checksum How to use checksums |
985d29c5405b4425c889e04dcc0773864712d003644f33af004e0766e399602d
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Release files / feagi_core-2.1.45-py3-none-any.whl
| Download URL | feagi_core-2.1.45-py3-none-any.whl |
|---|---|
| Size | 29.5 MB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
361c2b14e2c883b08efe5ef5140c3da1b381ef3a7ee08b6561a12dd3376614e9
|
|
BLAKE2b-256 checksum How to use checksums |
21bb69a75a62935a4a4fbc4e1e0969da828748fa91fbecbea258ae2bfb070d8e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Security for feagi-core 2.1.45
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
