Tool calling

Haystack

Open-source AI orchestration framework for production RAG pipelines and LLM agents.

LLM Agents

Minimal Python library for building tool-using agents controlled by language models

Mastra

TypeScript framework for building ai agents, workflows, and applications.

Model Context Protocol (MCP)

Open protocol for connecting AI applications with tools, data sources, and prompts.

OpenAI Agents SDK

Open-source SDK for building agents with tools, handoffs, guardrails, and tracing.

PydanticAI

Python framework for building typed ai agents with tools, structured outputs, and model integrations.

Qwen Agent

Open-source agent framework built on Qwen models with function calling, MCP, code interpreter, and RAG.

Smolagents

Open-source Python library for building and running model-agnostic agents with little code.

Arcade

MCP actions runtime that authenticates, executes, and governs agent tool calls.

Flowise AI

Open-source visual builder for multi-agent systems, chat assistants, and RAG.

Agent Tool Router JSON

Choose the next agent tool call from a user goal and available schemas, returning arguments in JSON.

Clothing Search Tool Call JSON

Convert a clothing request into a product-search call with category, subcategory, and color arguments.

Parallel Tool-Call Policy Block

Add a system-prompt block defining when an agent should run independent tool calls in parallel.

Add a parallel tool-call policy to an agent system prompt

Route a user goal to the next agent tool call as JSON

Turn a shopping request into a product search JSON call