Skip to content
Agent Search Engine.

System class · No. 02

Agent Frameworks

Open-source frameworks and SDKs for building, orchestrating, and deploying AI agents.

58 records · independent index · how we rank

Curated products

Alphabetical — stars don't measure commercial adoption

Repository-listed tools

Stars measure repository interest. A public repository does not establish an open-source license, product quality or active users; check each record and its official documentation.

→ Compare records marked open source in agent frameworks

By language: Python agent frameworks · TypeScript agent frameworks

Order this page by
langchainFrameworkThe agent engineering platform.Costs need recheckingOpen sourcePython · recorded license: MITRecord checked · Recent push 150kGitHub starsrufloFrameworkThe leading agent meta-harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems.Costs not checkedOpen sourceTypeScript · recorded license: MITRecord checked · Recent push 82kGitHub starsMetaGPTFrameworkA multi-agent framework that assigns software-company roles to LLM agents to turn a requirement into a project.Costs not checkedOpen sourcePython · recorded license: MITRecord checked 71kGitHub starsautogenFrameworkA programming framework for agentic AI; upstream is in maintenance mode.Costs not checkedOpen-source codeMaintenance modePython · recorded license: MITRecord checked · Recent push 62kGitHub starscrewAIFrameworkFramework for orchestrating role-playing, autonomous AI agents. By fostering collaborative intelligence, CrewAI empowers agents to work together seamlessly, tackling complex tasks.Costs need recheckingOpen sourcePython · recorded license: MITRecord checked · Recent push 61kGitHub starsLlamaIndexFrameworkFramework for connecting LLMs and agents to your own documents and data.Costs not checkedOpen sourcerecorded license: MITRecord checked · Recent push 53kGitHub starsLangGraphFrameworkBuild stateful workflows with code-defined and model-driven steps.Costs need recheckingOpen sourcePython · recorded license: MITRecord checked · Recent push 45kGitHub starsDSPyFrameworkStanford's framework for programming, not prompting, language models, with optimisers for prompts and weights.Costs not checkedOpen sourcePython · recorded license: MITRecord checked · Recent push 39kGitHub starsCopilotKitFrameworkThe frontend stack for agents: generative UI, shared state and human-in-the-loop for web, mobile, Slack and Teams.Costs not checkedOpen sourceTypeScript · recorded license: MITRecord checked · Recent push 39kGitHub starsAgentScopeFrameworkA production-ready Python agent framework built around model reasoning and tool use rather than rigid orchestration.Costs not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 36kGitHub starsOpenAI Agents SDK (Python)FrameworkOpenAI's lightweight, provider-agnostic Python framework for multi-agent workflows.Costs not checkedOpen sourcePython · recorded license: MITRecord checked · Recent push 31kGitHub starssmolagentsFrameworkMinimal agent library where agents write their actions as Python code.Costs need recheckingOpen sourcerecorded license: Apache-2.0Record checked · Recent push 30kGitHub stars
For agent buildersMake your product part of the discovery.Explore advertising
MastraFrameworkA TypeScript framework for AI applications and agents, with model routing, workflows, RAG and evals.Costs not checkedOpen sourceTypeScript · recorded license: Apache-2.0Record checked · Recent push 30kGitHub starsSemantic KernelFrameworkMicrosoft's model-agnostic SDK for AI agents in C#, Python and Java, now succeeded by Microsoft Agent Framework.Costs not checkedOpen sourceC# · recorded license: MITRecord checked · Recent push 29kGitHub starsAI SDK (Vercel)FrameworkVercel's provider-agnostic TypeScript toolkit for AI applications and agents in Next.js, React, Svelte, Vue and Node.Costs not checkedOpen sourceTypeScript · recorded license: Apache-2.0Record checked · Recent push 28kGitHub starsLetta (formerly MemGPT)FrameworkStateful agents and persistent memory tooling with Cloud, CLI and SDK interfaces.Code plus model/hosting costs; Cloud account and plan limitsOpen sourcerecorded license: Apache-2.0Record checked · Recent push 25kGitHub starsAgent Development Kit (Google ADK)FrameworkGoogle's open-source, code-first Python framework for building, evaluating and deploying AI agents.Costs not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 22kGitHub starspydantic-aiFrameworkAI Agent Framework, the Pydantic wayCosts need recheckingOpen sourcePython · recorded license: MITRecord checked · Recent push 21kGitHub starselizaFrameworkOpen source agentic operating systemCosts not checkedOpen sourceTypeScript · recorded license: MITRecord checked · Recent push 20kGitHub starsCAMELFrameworkAn open-source multi-agent framework from the CAMEL-AI community for studying and building agent societies.Costs not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 18kGitHub starsSuperAGIFramework<> SuperAGI - A dev-first open source autonomous AI agent framework. Enabling developers to build, manage & run useful autonomous agents quickly and reliably.Costs not checkedOpen sourcePython · recorded license: MITRecord checked 18kGitHub starsagent-frameworkFrameworkA framework for building, orchestrating and deploying AI agents and multi-agent workflows with support for Python and.NET.Costs not checkedOpen sourcePython · recorded license: MITRecord checked · Recent push 15kGitHub starstxtaiFrameworkAn all-in-one Python framework for semantic search, LLM orchestration and language-model workflows.Costs not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 13kGitHub starshiveFrameworkMulti-Agent Harness for Production AICosts not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 11kGitHub starsVoltAgentFrameworkAn open-source TypeScript agent framework with a companion console for observability, evals and deployment.Costs not checkedOpen sourceTypeScript · recorded license: MITRecord checked · Recent push 11kGitHub starsharness-sdkSDK / libraryBuild an agent harness and control it end-to-end. Open-source SDK for production AI agents in Python & TypeScript - any model, any cloud.Costs not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 9.2kGitHub starsAgent Development Kit for Go (Google ADK)FrameworkGoogle's open-source, code-first Go toolkit for building, evaluating and deploying AI agents.Costs not checkedOpen sourceGo · recorded license: Apache-2.0Record checked · Recent push 9.0kGitHub starsevolverFrameworkThe GEP-powered self-evolving engine for AI agents. Auditable evolution with Genes, Capsules, and Events. | evomap.aiCosts not checkedOpen sourceJavaScript · recorded license: GPL-3.0Record checked · Recent push 8.7kGitHub starsmcp-agentFrameworkA composable Python framework for building agents from MCP servers and simple workflow patterns.Costs not checkedOpen sourcePython · recorded license: Apache-2.0Record checked 8.6kGitHub starsyaoFrameworkBuild AI agents and web apps — with a single binary.Costs not checkedLicense not checkedGoRecord checked · Recent push 8.3kGitHub starsUpsonicFrameworkBuild autonomous AI agents in Python.Costs not checkedOpen sourcePython · recorded license: MITRecord checked 8.0kGitHub starsSwarmsFrameworkA Python multi-agent orchestration framework with prebuilt sequential, concurrent and hierarchical architectures.Costs not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 7.4kGitHub starsopen-multi-agentFrameworkTypeScript multi-agent orchestration framework. Describe a goal, a coordinator decomposes it into a task DAG that runs on any LLM: Claude, ChatGPT, Gemini, DeepSeek, or local models.Costs not checkedOpen sourceTypeScript · recorded license: MITRecord checked · Recent push 7.1kGitHub starsagent-governance-toolkitSDK / libraryAI Agent Governance Toolkit — Policy enforcement, zero-trust identity, execution sandboxing, and reliability engineering for autonomous AI agents. Covers 10/10 OWASP Agentic Top 10.Costs not checkedOpen sourcePython · recorded license: MITRecord checked · Recent push 6.6kGitHub starsintentkitFrameworkIntentKit is an open-source, self-hosted cloud agent cluster that manages a collaborative team of AI agents for you.Costs not checkedOpen sourcePython · recorded license: MITRecord checked 6.5kGitHub starsAtomic AgentsFrameworkA lightweight, modular Python framework for building agent pipelines from small, composable pieces.Costs not checkedOpen sourcePython · recorded license: MITRecord checked · Recent push 6.3kGitHub starsagentsFrameworkAn Open-source Framework for Data-centric, Self-evolving Autonomous Language AgentsCosts not checkedOpen sourcePython · recorded license: Apache-2.0Record checked 6.0kGitHub starsrllmFrameworkDemocratizing Reinforcement Learning for LLMsCosts not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 5.9kGitHub starsag2FrameworkAG2 (formerly AutoGen): The Open-Source AgentOS.Join us at: https://discord.gg/sNGSwQME3xCosts not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 5.1kGitHub starsKoogFrameworkJetBrains' Kotlin and Java framework for predictable, fault-tolerant AI agents across JVM, Android, iOS and the browser.Costs not checkedOpen sourceKotlin · recorded license: Apache-2.0Record checked · Recent push 4.7kGitHub starsyoutu-agentFrameworkA simple yet powerful agent framework that delivers with open-source modelsCosts not checkedLicense not checkedPythonRecord checked · Recent push 4.6kGitHub starsLangroidFrameworkA lightweight Python framework for multi-agent LLM applications, from CMU and UW-Madison researchers.Costs not checkedOpen sourcePython · recorded license: MITRecord checked · Recent push 4.2kGitHub starsOpenAI Agents SDK (TypeScript)FrameworkOpenAI's lightweight, provider-agnostic TypeScript framework for multi-agent workflows and voice agents.Costs not checkedOpen sourceTypeScript · recorded license: MITRecord checked · Recent push 4.0kGitHub starsfast-agentFrameworkCode, Build and Evaluate agents - excellent Model and Skills/MCP/ACP/A2A SupportCosts not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 4.0kGitHub starsVibe-SkillsSDK / libraryVibe-Skills is an all-in-one AI skills package.Costs not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 3.8kGitHub starsBeeAI FrameworkFrameworkA framework for production-ready multi-agent systems with matching Python and TypeScript libraries.Costs not checkedOpen sourcePython · recorded license: Apache-2.0Record checked · Recent push 3.5kGitHub stars
How to choose agent frameworks

Agent frameworks are the libraries and SDKs you build on to orchestrate LLM calls, tools, memory, and multi-agent workflows — the scaffolding between a raw model API and a working agent. The decision usually hinges on your language, how much control versus abstraction you want, and whether you need production features like observability, retries, durable execution, and human-in-the-loop checkpoints. Lightweight libraries give you fine-grained control and stay out of the way but leave more to wire up yourself; higher-level frameworks move faster at the cost of flexibility and can be harder to debug when an abstraction leaks. Check first-class support for your model providers and for tool and function calling, then look at how state and memory are handled and whether streaming works end to end.

Most frameworks are open source and free; many companies pair an open-source core with a paid hosted platform for tracing, deployment, and scale, so factor in where this will actually run in production. In a field moving this fast, ecosystem and momentum matter more than feature checklists — a framework with active maintainers, real adoption, and solid docs will save you more time than one with a longer feature list and a stalled repository. Prototype your hardest workflow in two candidates before committing; the one that fights you least is usually the right answer.

Building one yourself? How to choose an agent framework →

Frequently asked

What is an AI agent framework?
A framework provides the building blocks — tool calling, memory, planning, and multi-agent orchestration — for constructing agents in code, instead of wiring everything to an LLM API by hand.
Which agent framework should I use?
Match it to your stack and control needs: lightweight libraries give fine-grained control, higher-level platforms trade flexibility for speed. Check language support, production tooling, and how active the project is.
Are agent frameworks open source?
Most are. Many companies pair an open-source framework with a paid hosted platform for deployment, tracing, and scaling.
Agent Frameworks — directory | Agent Search Engine