CrewAI
Framework for orchestrating role-playing, autonomous agents that collaborate on tasks.
Website captured Oct 19
Sub-agents & orchestration
Split work across agents with clean, isolated contexts: agent frameworks, sub-agents, handoffs, workflows.
17 listed entries
tool
LangGraph
Framework for stateful, long-running agents built as graphs, with checkpoints and human-in-the-loop.
Website captured Oct 19
Agno
Framework and runtime for building, running and managing multi-agent systems.
Website captured Oct 19
DSPy
Framework for programming, not prompting, language models: modules and optimizers that compile prompts.
Website captured Oct 19
OpenAI Agents SDK
Lightweight framework for multi-agent workflows with handoffs, guardrails and tracing.
Website captured Oct 19
smolagents
Hugging Face's barebones library for agents that think and act in code.
Website captured Oct 19
Mastra
TypeScript framework for AI agents, workflows, RAG and memory.
Website captured Oct 19
AI SDK (Vercel)
TypeScript toolkit for AI apps and agents: tool calling, streaming and structured output across providers.
Website captured Oct 19
Agent Development Kit (ADK)
Google's code-first toolkit for building, evaluating and deploying agents (Python, with a JS SDK).
Website captured Oct 19
Pydantic AI
Type-safe Python agent framework from the Pydantic team, working across model providers.
Website captured Oct 19
Microsoft Agent Framework
Microsoft's framework (Python and .NET) for building, orchestrating and deploying agents and multi-agent workflows.
Website captured Oct 19
BAML
A programming language for agents: typed prompt functions with schema-aligned parsing of model output.
Website captured Oct 19
Claude Agent SDK
Build agents on Claude Code's harness from code: tools, sub-agents, context management and permissions.
Website captured Oct 19
Context engineering
The set of strategies for curating and maintaining the optimal set of tokens during LLM inference, including everything that lands in context outside the prompt.
Keep errors in context
Leave failed actions and error messages in the history so the model learns from them instead of repeating them.
Sub-agent isolation
Give focused tasks to sub-agents with clean context windows; they return condensed summaries to a lead agent.
Write, select, compress, isolate
LangChain's four buckets of context engineering: save context outside the window, pull the right parts in, shrink it, or split it across agents and environments.
How they connect
- Claude Agent SDK alternative to OpenAI Agents SDK
- LangGraph alternative to Microsoft Agent Framework
- LangGraph alternative to CrewAI
- LangGraph alternative to OpenAI Agents SDK
- Microsoft Agent Framework alternative to Agent Development Kit (ADK)
- CrewAI alternative to Agno
- DSPy alternative to BAML
- Sub-agent isolation implements Claude Agent SDK
- Sub-agent isolation implements LangGraph
- Sub-agent isolation implements OpenAI Agents SDK
- Mastra alternative to LangGraph
- Mastra alternative to AI SDK (Vercel)
- smolagents alternative to Pydantic AI
- Pydantic AI alternative to Agno
- OpenAI Agents SDK alternative to Agent Development Kit (ADK)