From a goal to a task DAG, automatically. TypeScript-native multi-agent orchestration with MCP and live tracing. Three runtime dependencies.

Awesome MCP Servers - A curated list of Model Context Protocol servers

🎨 NeMo Data Designer: Generate high-quality synthetic data from scratch or from seed data.

ReSearch: Learning to Reason with Search for LLMs via Reinforcement Learning & ReCall: Learning to Reason with Tool Call for LLMs via Reinforcement Learning

Democratizing AI scientists with ToolUniverse

An autonomous agent that takes work, does work, gets paid, and gets better at it.

A coding agent optimized to smaller LLMs

LLaVA-Plus: Large Language and Vision Assistants that Plug and Learn to Use Skills

Open-source agent platform for Global × China enterprises — wire every system through one agent core. Self-hosted, any LLM.

Easily create LLM tools and agents using plain Bash/JavaScript/Python functions.

"AnyTool: Universal Tool-Use Layer for AI Agents"

Official repository for DR Tulu: Reinforcement Learning with Evolving Rubrics for Deep Research

TxAgent: An AI Agent for Therapeutic Reasoning Across a Universe of Tools

kani (カニ) is a highly hackable microframework for tool-calling language models. (NLP-OSS @ EMNLP 2023)

Simple to install, powerful command-line based AI agent system for coding.

ScienceClaw is a personal research assistant built with LangChain DeepAgents and AIO Sandbox infrastructure, adopting a completely new architecture beyond OpenClaw. It offers stronger security, better transparency, and a more user-friendly experience.

Second Brain is an agentic framework that acts as an operating system, using local file intelligence, workflow automation, and LLMs to complete tasks and communicate over multiple modalities and messaging platforms.

[ICLR'25] BigCodeBench: Benchmarking Code Generation Towards AGI

MCPMark is a comprehensive, stress-testing MCP benchmark designed to evaluate model and agent capabilities in real-world MCP use.

DeepSeek-native AI coding agent for your terminal. Engineered around prefix-cache stability — leave it running.