All Frameworks
26 agent frameworks and SDKs
Full-featured framework for building LLM applications with chains, agents, and retrieval. Supports tool calling, structured output, memory, and streaming across dozens of model providers.
pip install langchain langchain-openaiFramework for building AI agents that can browse the web autonomously. Handles navigation, form filling, clicking, and data extraction with vision or DOM-based interaction.
pip install browser-useMicrosoft's framework for building multi-agent conversational systems. Agents converse with each other and humans to solve tasks through flexible conversation patterns.
pip install autogen-agentchatMicrosoft Research's multi-agent system with specialized agents for web browsing, file handling, and code execution. Orchestrated by a lead agent that delegates to specialists.
pip install autogen-ext[magentic-one]Data framework for connecting LLMs to external data. Specializes in ingestion, indexing, and retrieval with advanced RAG patterns, query engines, and agent workflows.
pip install llama-indexAnthropic's SDK for building custom agents powered by Claude. Provides programmatic access to Claude's agentic capabilities including tool use, computer use, and extended thinking.
npm install @anthropic-ai/claude-codeRole-playing multi-agent framework where you define agents with specific roles, goals, and backstories. Agents collaborate on tasks with configurable delegation and process flows.
pip install crewaiMicrosoft's enterprise-grade SDK for integrating LLMs into applications. Offers plugins, planners, and memory with first-class .NET and Python support.
pip install semantic-kernelProgramming framework that compiles declarative language model calls into optimized prompts or finetunes. Replaces hand-written prompts with modular, trainable programs.
pip install dspyFull-stack framework for building, shipping, and monitoring agentic systems. Supports multi-modal agents with reasoning, knowledge, memory, and team coordination.
pip install agnoLightweight multi-agent orchestration framework from OpenAI. Explores ergonomic patterns for handoffs and routines with minimal abstraction over the Chat Completions API.
pip install git+https://github.com/openai/swarm.gitProduction-ready framework for building RAG pipelines, semantic search, and question answering systems. Component-based architecture with strong focus on retrieval quality.
pip install haystack-aiFramework for building agentic systems with memory, knowledge, and tools. Supports teams of agents with shared state, built-in web search, and playground UI.
pip install phidataOfficial OpenAI SDK for building agents with tool use, handoffs, and guardrails. Lightweight, opinionated design with native support for OpenAI models and the Responses API.
pip install openai-agentsTool integration platform for AI agents. Provides 250+ pre-built tool integrations (GitHub, Slack, Jira, etc.) with managed auth, compatible with any agent framework.
pip install composio-coreHugging Face's minimal agent library focused on code-first agent design. Agents write and execute Python code to accomplish tasks rather than using JSON tool calls.
pip install smolagentsFramework for building agents with long-term memory and self-editing capabilities. Implements virtual context management so agents can maintain persistent state across conversations.
pip install lettaTypeScript toolkit for building AI-powered applications with React, Next.js, and other frameworks. Unified provider interface, streaming UI components, and tool calling with structured output.
npm install ai @ai-sdk/openaiTypeScript-first agent framework with built-in tool calling, RAG, workflows, and integrations. Designed for full-stack TypeScript developers who want agents without leaving their stack.
npm install mastraStructured extraction library that patches LLM clients to return Pydantic models. Simple API for reliable structured output with automatic retries and validation.
pip install instructorAll-in-one embeddings database for semantic search, LLM orchestration, and language model workflows. Combines vector search with RAG pipelines in a single library.
pip install txtaiGoogle's framework for building AI agents with Gemini models. Supports multi-agent systems, tool use, session management, and deployment to Google Cloud with built-in evaluation.
pip install google-adkType-safe agent framework built on Pydantic. Brings data validation to LLM interactions with structured responses, dependency injection, and model-agnostic design.
pip install pydantic-aiGraph-based orchestration framework for building stateful, multi-actor LLM applications. Defines agent workflows as directed graphs with conditional edges, cycles, and persistence.
pip install langgraphResearch-oriented multi-agent framework for studying cooperative AI agent behavior. Implements role-playing, task decomposition, and society simulation for LLM agents.
pip install camel-aiGoogle's open protocol for inter-agent communication. Defines how AI agents discover each other, negotiate capabilities, and collaborate on tasks across different frameworks.
pip install a2a-sdk