Foundational Python framework for building LLM-powered apps, chatbots, and agents with modular components for prompt templating, memory, tools, and orchestration. 47M+ PyPI downloads. Note: No public affiliate program; direct link tracked.
State-machine agent runtime built by LangChain. Models workflows as directed graphs with checkpointing, human-in-the-loop, and streaming. v1.0 released late 2025.
Production-grade Python SDK replacing OpenAI Swarm (deprecated). Supports sequential, hierarchical, and consensual agent processes. Provider-agnostic with 100+ LLM connectors.
TypeScript-native AI framework for building agents, RAG pipelines, and multi-agent workflows. First-class TypeScript support — no translation layer needed.
Python-first agent framework where agent outputs are guaranteed to conform to Pydantic schemas. Best for developers who want type safety + LLM flexibility.
Data framework with first-class agent support — 160+ data connectors for RAG-grounded agents. 47K GitHub stars. Excellent for data-intensive use cases.
Official Anthropic SDK powering Claude Code. TypeScript + Python, supports hooks, MCP, skills, subagents, and streaming. Best for Anthropic-native production agents.
Declarative AI programming framework. Define what agents should do via signatures rather than prompt strings. Compiles to optimized prompts.
Deepset's open-source framework for building LLM applications and RAG pipelines with agentic capabilities. Strong in European enterprise.