HybridAGI is the first Programmable LLM-based Autonomous Agent that lets you program its behavior using a graph-based prompt programming approach. This state-of-the-art feature allows the AGI to efficiently use any tool while controlling the long-term behavior of the agent.
-
Efficient Storage: Thanks to an hybrid vector and graph database powered by FalkorDB enabling efficient storage of data you can inspect its long-term memory and known what's going on in a glance!
-
Graph-based Prompt Programming: HybridAGI allows you to encode its behavior using programs represented as graphs. This capability, at the core of our approach, ensures that the system follows a structured and logical behavior. Want to adapt its behavior to your workflow? Learn how to program HybridAGI using Cypher!
-
Graph Program Interpreter: We introduce a revolutionary Agent that leverages probabilistic decision making and graphs to determine actions based on a program. By reducing ambiguity and allowing composition of programs, this state-of-the-art feature enables the AGI to handle complex tasks with ease and precision.
-
Free Software: HybridAGI is a community-driven project, fostering collaboration, innovation, and shared ownership. The software is released under the GNU GPL license, inviting contributions from a diverse range of users and empowering the collective intelligence of the community. Its architecture allows you to release your Cypher programs under the license of your choice while using the framework under GNU GPL.
First, clone the chat repository with:
git clone https://github.com/SynaLinks/HybridAGI-chat
cd HybridAGI-chatThen you should open the repository folder in your favorite IDE (VSCodium with the Neo4J plugin is a good start).
📦HybridAGI-chat
┣ 📂archives # This is where the AGI will save its uploaded work
┣ 📂documentation # This is where you can put your pdf and documents
┣ 📂programs # This is where you should put your Cypher programs
┣ 📂src # The source code of the UI
... the license and other files related to deploymentNote that the folders archives, documentation and programs are shared with the application container, you can edit them and reload your programs/documentation without restarting the application container.
Start with a simple echo test, create a main.cypher file inside the programs folder:
// Nodes declaration
CREATE
(start:Control {name:"Start"}),
(end:Control {name:"End"}),
(echo_objective:Action {
name:"Reformulate the Objective",
tool:"Speak",
prompt:"Please reformulate the objective using other words"}),
// Structure declaration
(start)-[:NEXT]->(echo_objective),
(echo_objective)-[:NEXT]->(end)Learn more about Graph-based Prompt Programming by reading our documentation.
Now it is time to deploy this app, just use the following command
docker-compose upOpen your browser at http://localhost:8001 and connect to an existing database with the hostname falkordb and port 6379.
HybridAGI is made possible by the following open-source tools:
If you found this repository usefull for your research, please consider citing us:
@misc{Sallami2023,
author = {Yoan Sallami},
title = {HybridAGI: Introducing graph-based prompt programming},
year = {2023},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/SynaLinks/HybridAGI}}
}Become a part of our community of developers, researchers, and AI enthusiasts. Contribute to the project, share your feedback, and help shape the future of Hybrid AGI. We welcome and value your participation!
We're a small team dedicated to advancing the future of AI through open-source research. Consider supporting our work by contributing the cost of a coffee or by sharing this repository on social media! Your support helps us continue our research and development. Additionally, feel free to check out our progress, findings, and code on GitHub. We're grateful for any support and collaboration from the community as we work towards shaping the future of AI. Thank you!