Open-source LLM app development platform combining visual agentic workflows, RAG pipelines, model management and observability in one workspace.
neuron-core/neuron-ai
A PHP framework for building agentic AI applications, connecting LLM providers, tools, MCP servers, RAG and workflows to agents.
- Stars
- 2,152
- Forks
- 257
- Open issues
- 10
- Last commit
- yesterday
What is it?
Neuron is a PHP framework for creating and orchestrating AI agents inside PHP applications. An agent is a PHP class that extends the framework's Agent base class, defines its provider and instructions, and can be given tools, memory and RAG capabilities. Conversation history is bound to a thread ID, so an agent can keep context across messages.
The framework lets you switch between many LLM providers, including Anthropic, OpenAI, Gemini, Mistral, Ollama and AWS Bedrock, without changing agent code. Tools and toolkits let agents call PHP functions, such as a MySQL schema and select toolkit, and an MCP connector can load tools exposed by external MCP servers. Structured output maps model responses onto PHP classes annotated with schema attributes.
For retrieval use cases, a RAG base class combines an embeddings provider with a vector store such as Pinecone. A Workflow component exposes the underlying pieces as standalone building blocks and supports human-in-the-loop pauses, and the project includes Laravel and Symfony integration skills and a console command generator.
Key features
- 01Switch between many LLM providers without changing agent code
- 02Give agents tools, toolkits and MCP server tools
- 03Structured output mapped to PHP classes via attributes
- 04Built-in RAG with embeddings providers and vector stores like Pinecone
- 05Workflow engine with human-in-the-loop pauses and checkpointing
Use cases
- Adding a conversational assistant to an existing Laravel or Symfony app
- Letting an agent answer questions by querying a MySQL database read-only
- Extracting structured data from natural language into PHP objects
- Building a retrieval-augmented chatbot over documents stored in a vector database
Quick start
- 1
Install the package with Composer
composer require neuron-core/neuron-ai - 2
Generate a starter agent class
vendor/bin/neuron make:agent DataAnalystAgent - 3
Set your provider API key and model in the agent, then chat with it
Things to know
- Requires PHP 8.1 or newer.
- Most examples call a paid LLM provider API, so you need an account and API key.
In collections
Topics
FAQ
- What is neuron-ai?
- A PHP framework for building agentic AI applications, connecting LLM providers, tools, MCP servers, RAG and workflows to agents. Neuron is a PHP framework for creating and orchestrating AI agents inside PHP applications. An agent is a PHP class that extends the framework's Agent base class, defines its provider and instructions, and can be given tools, memory and RAG capabilities. Conversation history is bound to a thread ID, so an agent can keep context across messages.
- Is neuron-ai free to use? What license does it have?
- neuron-ai is open source under the MIT license. Check the license terms before using it in commercial products.
- Is neuron-ai still maintained?
- Actively maintained. The last commit to neuron-ai was on Oct 21, 2026.
- What can I use neuron-ai for?
- Adding a conversational assistant to an existing Laravel or Symfony app Letting an agent answer questions by querying a MySQL database read-only Extracting structured data from natural language into PHP objects Building a retrieval-augmented chatbot over documents stored in a vector database
- How do I install neuron-ai?
- Install the package with Composer: composer require neuron-core/neuron-ai Generate a starter agent class: vendor/bin/neuron make:agent DataAnalystAgent Set your provider API key and model in the agent, then chat with it
- Where can I get neuron-ai?
- The source code and installation instructions are on GitHub: https://github.com/neuron-core/neuron-ai
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A zero-code platform that generates production-style AI agents from natural-language descriptions, with built-in tools, memory and orchestration.
A multi-agent system built on Context Engineering, with a visual editor for designing, testing and deploying agent workflows.
Open-source platform for building enterprise-grade AI agents, with RAG knowledge bases, visual workflows and MCP tool use.
Open-source AI app builder for creating and running agentic workflows with a visual drag-and-drop editor and multiple LLM providers.
A local-first, Markdown-native long-term memory layer for AI agents, storing memories as editable files with SQLite and LanceDB indexes.