langchain-ai/langchain
LangChain is a versatile framework for building and deploying applications powered by large language models, offering interoperability, real-time data integration, and advanced agent orchestration capabilities.
Awesome AI Agents › Language Agent Frameworks
howl-anderson/agentsilex
AgentSilex is a Python agent framework built around a deliberately small core of roughly three hundred lines, so a developer can read the whole implementation before depending on it. The library exposes four primitives: an Agent that holds a name, a model identifier, instructions and a list of tools; a tool decorator that turns an ordinary Python function into a callable tool and derives its schema automatically; a Session that stores conversation history; and a Runner that drives the loop of sending messages to the model, executing requested tool calls and feeding the results back. Model access goes through LiteLLM, so the same agent code runs against OpenAI, Anthropic, Google Gemini, DeepSeek, Azure, Mistral and local endpoints by changing a single model string. Beyond single agents, the framework supports handoffs: a coordinating agent lists other agents in its handoffs argument and transfers a conversation to a specialist, which gives a simple form of multi-agent routing without a separate orchestration layer. Responses can be streamed as they arrive, and structured output is supported through Pydantic models. Observability is built in through OpenTelemetry tracing, and the documentation shows how to point traces at Arize Phoenix to inspect a tree of agent turns, tool calls and handoffs. The project positions itself against larger frameworks whose control flow is hidden from the user, arguing that a framework you can read end to end is easier to customise and to debug. Its stated audiences are companies that want a foundation they can modify, developers who want to understand agent internals rather than only call them, educators teaching how agents work, and researchers prototyping new agent architectures. It is published on PyPI as agentsilex, requires Python 3.12 or newer, and is released under the MIT licence.
https://github.com/howl-anderson/agentsilex
Also in Language Agent Frameworks
langchain-ai/langgraph
LangGraph is a low-level orchestration framework for building resilient, controllable AI language agents with customizable architectures, long-term memory, and real-time streaming support, used by major companies and integrated with the LangChain ecosystem.
microsoft/semantic-kernel
Semantic Kernel is an enterprise-ready orchestration framework by Microsoft that enables developers to build and deploy intelligent AI agents and multi-agent systems using cutting-edge large language model technology.
dataelement/bisheng
BISHENG is an open-source LLM DevOps platform designed for enterprise AI applications, offering powerful workflow orchestration, security features, and high-precision document parsing to support complex intelligent business solutions.
VoltAgent/voltagent
VoltAgent is an open-source TypeScript framework for building and orchestrating AI agents powered by Large Language Models, enabling developers to create scalable, customizable, and maintainable AI applications with modular components and visual monitoring.
kyegomez/swarms
Swarms is an enterprise-grade, production-ready multi-agent orchestration framework designed to enable complex AI workflows with high scalability, flexibility, and robust developer and security features.
langchain-ai/langgraphjs
LangGraph.js is a low-level orchestration framework for building customizable, reliable, and controllable language agents as graphs, enabling complex AI workflows with features like long-term memory, human-in-the-loop, and real-time streaming support.
neuron-core/neuron-ai
PHP framework for creating and orchestrating AI agents, covering LLM providers, tools, RAG, MCP, event-driven workflows with checkpointing and human-in-the-loop, streaming UI protocols and monitoring.