New in week 43

Python framework for multi-agent teams and workflows, with a CLI and Docker, MCP and A2A deployment options.

Open-source access layer that connects AI agents to 1200+ integrations with shared auth, permissions and audit logs.

Open framework for building and hosting production-grade AI agents and multi-agent workflows in .NET and Python.

Policy enforcement, identity, sandboxing and reliability tooling for autonomous AI agents, installable with one pip command.

Reinforcement-learning trainer of about 3,500 lines that trains models behind agents run in their own harnesses, locally or as Kubernetes jobs.

SDK for building agents in C#, JavaScript and Python that run on channels such as Microsoft Teams and Microsoft 365 Copilot.

Archived Microsoft Python framework for multi-agent apps; Microsoft directs new users to Microsoft Agent Framework.

Visualization language and MCP server that lets AI agents create charts from editable chart specs.

Model-agnostic SDK for building agents and multi-agent systems in Python, .NET and Java; Microsoft names Agent Framework as its successor.

Java SDK for calling LLMs such as OpenAI, Azure OpenAI and Hugging Face, with plugins that chain functions and planners.

SDK for building Microsoft Teams apps and AI agents in TypeScript, C# and Python, with a CLI that scaffolds new agents.

Python and TypeScript toolkit for calling any frontier LLM through one interface, with tools, structured output and tracing.

Reinforcement learning framework for training reasoning LLM agents over multi-turn episodes with the StarPO algorithm.

Lightweight Python agent framework for long-running tasks, with skills, sub-agents, A2A and ACP servers, and CLI, TUI and web front ends.

Python interface for running and evaluating agents from several frameworks, with OpenTelemetry traces; now in soft deprecation.

All-in-one Python AI framework for embeddings-based semantic search, LLM pipelines, workflows and agents.

PHP framework for agentic applications with workflows, human-in-the-loop, MCP connectors, RAG and multi-provider LLM support.

Python SDK for TimeGPT, a time series foundation model for forecasting and anomaly detection.

Browser infrastructure for AI agents: local and cloud Playwright-based browser sessions, scraping and deployable browser functions.

Open-source NVIDIA library that connects existing agents to tools and data, with profiling and LangSmith tracing; deprecated, EOL end of 2026.

OpenAI-compatible agentic API server for inference, vector stores, files and skills, on any model or infrastructure.

Self-hosted TypeScript runtime for coordinating agents with human approval gates and inspectable runs, with no telemetry.

TypeScript and JavaScript framework for multi-agent workflows with tools, handoffs, guardrails, sandboxes and voice agents.

Python SDK for multi-agent workflows with handoffs, guardrails, sessions and built-in tracing, usable with OpenAI and other LLMs.

Experimental Python library that teaches agent handoffs on the Chat Completions API; OpenAI now points users to its Agents SDK.

Python and TypeScript SDK and REST Agent Server for building agents that work with code in local or containerized workspaces.

Open-source reinforcement learning framework that trains LLM agents on multi-step tasks using GRPO.

Embeddable version control system for files and SQL tables, with branch, diff, merge and rollback across both.

Python and TypeScript SDK that gives an AI agent a phone number, with pluggable LLM, speech and telephony providers.

Ruby library for LLM-powered apps with a unified LLM interface, prompt tools, output parsers, RAG and assistants.

Open-source Python framework for real-time voice and multimodal agents, built from composable pipelines with handoffs between specialist agents.

Python framework for structured LLM outputs and agentic workflows built from tasks, agents and threads.

Python framework for agents that discover MCP tools by themselves, with optional memory, guardrails, cache and tracing.

Typed Python SDK for building agents with swappable model providers, plus optional capabilities for memory, sub-agents, planning and durable execution.

Quarkus extensions that connect LangChain4j to Quarkus apps, with declarative AI services and LLM, tool and embedding support.

Python framework for LLM applications built on Qwen models, with function calling, MCP, a code interpreter and RAG options.

Open-source Python framework for text and voice assistants; the open-source edition is in maintenance mode while Rasa moves to its CALM engine.

PyTorch-style Python framework for multi-agent, multi-session workflows, with a durable production runtime in v2.0.0.

Open-source Python toolkit for RAG and agents over your own data, with event-driven workflows and 300+ integration packages.

C# framework on .NET for building multi-agent AI bots with plugins, RAG, MCP integration and a built-in admin UI.

Framework for developing, evaluating and benchmarking web agents on BrowserGym benchmarks, with parallel runs through Ray.

Gym-style Python environment for building and evaluating web agents on MiniWoB, WebArena, WorkArena and other benchmarks.

Java application framework from the Spring project with portable chat and embedding APIs, tool calling and vector stores.

Python framework for writing language-model programs as code and optimizing their prompts and weights instead of hand-tuning prompts.

XState library where agent logic is a state machine, so a model can only pick events the machine allows.

Open-source Python and TypeScript SDK for model-driven agents, with tools, MCP, multi-agent patterns and streaming.

Sample projects for the Strands Agents SDK, showing how to build agents in Python and TypeScript; not intended for production use.

Browser DOCX editor and document engine built on OOXML, with a Document API that agents reach through SDKs, a CLI and an MCP server.

TypeScript SDK for streaming chat, tool-calling agents and structured output across model providers and UI frameworks.

Open-source framework for real-time multimodal conversational AI, with RTC and WebSocket voice assistant examples.