Open Source LLM Frameworks
Large language model (LLM) frameworks help developers build applications that use models to understand and generate text, images, or other content. They provide reusable components for connecting to models, structuring prompts, managing context, calling tools, and coordinating multi-step workflows. These capabilities reduce the effort of integrating model APIs and help make application behavior easier to develop, test, and maintain.
Open source tools in this area range from lightweight libraries for prompt construction to frameworks for agents, multimodal applications, and multi-agent workflows. When choosing one, consider its license, documentation, maintenance activity, supported models, runtime requirements, and compatibility with your existing stack. Also assess how easily you can test, observe, and control model behavior. These frameworks can be useful to software developers, researchers, and teams building AI-powered applications.
5 repositories · updated September 7, 2026
Best of Agent Harnesses: A Curated List for AI Agent Development
RyanAlberts' Best of Agent Harnesses is a comprehensive, curated, and ranked list of over 100 AI agent harnesses and orchestration frameworks. It provides valuable insights for building reliable agentic systems, offering both human-readable guides and machine-readable formats for agents themselves. The repository is rescored weekly to ensure up-to-date recommendations.
EasyInstruct: An Easy-to-Use Instruction Processing Framework for LLMs
EasyInstruct is an open-source Python framework designed to simplify instruction processing for Large Language Models (LLMs). Accepted at ACL 2024, it offers modularized components for instruction generation, selection, and prompting, supporting various LLMs like GPT-4 and LLaMA. This framework is ideal for researchers and developers working on LLM-based experiments and applications.
GenAIScript: Automatable GenAI Scripting with TypeScript/JavaScript
GenAIScript is an open-source project from Microsoft that enables programmatic assembly of prompts for Large Language Models (LLMs) using JavaScript and TypeScript. It allows developers to orchestrate LLMs, tools, and data directly in code, streamlining the development of GenAI applications. This framework offers seamless Visual Studio Code integration and a flexible command-line interface for efficient GenAI scripting.
fast-agent: Build and Orchestrate Multimodal AI Agents and Workflows
fast-agent is a powerful Python framework designed for creating and interacting with sophisticated multimodal AI agents and workflows. It offers a simple, declarative syntax for defining agents, comprehensive model support, and unique features like end-to-end tested MCP (Multi-modal Communication Protocol) integration. Developers can rapidly build, test, and deploy complex agent applications with advanced capabilities such as structured outputs, vision, and various orchestration patterns.
ROMA: Build Hierarchical Multi-Agent Systems
ROMA is a Python framework for solving complex tasks through recursive planning and coordinated agents. It suits developers building extensible multi-agent workflows with LLMs, tools, and optional persistence or API services.