The article explains that while an Agent’s core loop is only a few lines of code, the real engineering challenges lie in prompt design, context management, tool selection, and safety checks that together determine the loop’s effectiveness.
The article defines AI Agent as a system centered on a large model that can invoke tools, receive feedback, and continuously accomplish tasks, and systematically distinguishes related terms such as Model, Scaffolding, Harness, Context Engineering, Policy, Tool, Skill, Sub‑agent, Environment, Rollout, Reward, and Trainer, using concrete examples to clarify each.
This article breaks down the full technology stack for production‑ready AI agents in 2026, covering model gateways, orchestration frameworks, tool‑use protocols, memory layers, state‑machine execution, sandboxing, observability, evaluation, and human‑in‑the‑loop safeguards, while highlighting concrete tools, risks, and best‑practice trade‑offs.
This article breaks down the core terminology of AI agents—Model, Scaffold, Harness, Context Engineering, Policy, Tool Use, Skills, Sub‑agents, and the training‑side concepts of RL Environment, Trainer, Rollout, and Reward—explaining their roles, differences, and how they combine to form functional agents.
This article provides a comprehensive overview of the "Code as Agent Harness" paradigm, detailing its three‑layer architecture, the roles of code in reasoning, acting, and environment modeling, the mechanisms that enable reliable long‑term execution, and how multi‑agent systems scale the harness through shared code and feedback loops.
The article introduces the ICML 2026 Theory of Agent (ToA), analyzes four common failure modes of modern agents, explains the internal‑vs‑external tool trade‑off through a knowledge‑boundary framework, and outlines how effort‑conservation and the β parameter guide self‑evolving agent design and future research.
An AI Agent combines a large language model with a clear goal, callable tools, and a multi‑step reasoning loop, enabling perception, planning, and action that go beyond simple chat by decomposing tasks, using external APIs, iterating on errors, and managing memory, while acknowledging its limitations.
The article analyzes why traditional Retrieval‑Augmented Generation fails on multi‑hop, vague, or multi‑source queries and explains how Agentic RAG uses an LLM‑driven agent loop to make dynamic retrieval decisions, outlining its architecture, suitable scenarios, and limitations.
The article analyzes MiniMax’s strategy of co‑evolving large language models with a Harness framework, contrasting product philosophies, detailing a live MaxHermes demo that creates and refines reusable Skills, and explaining how this dual evolution reshapes the competitive focus from single‑turn Q&A to sustained, self‑improving agent workflows.
The article explains what an AI Agent is, how it differs from a large language model, outlines its three core abilities—autonomous planning, tool use, and memory—shows a step‑by‑step example, and discusses why agents have become popular and where they can be applied.
This article breaks down Microsoft’s open‑source AI‑Agent learning repository, explaining core concepts, five design patterns, production deployment considerations, and emerging protocols, while offering practical engineering guidance for building reliable multi‑agent systems from scratch.
This article dissects the leaked Claude Code v2.1.88 source to reveal how the three core concepts—Rules, MCP, and Skills—are implemented, where they are injected in the Anthropic LLM API request, and when developers should prefer each mechanism for reliable, secure, and token‑efficient coding agent workflows.
In this detailed lecture, Prof. Li Hongyi of National Taiwan University dissects the OpenClaw AI Agent, explaining its system prompts, tool usage, memory handling, sub‑agents, security risks like prompt injection, and practical safeguards for deploying autonomous agents on personal computers.
This article walks developers through the learn-claude-code project, teaching them how to construct a Claude‑style AI Agent Harness by covering twelve progressive lessons, core concepts such as agents, harnesses, sub‑agents, context compression, task management, and providing runnable Python examples and architectural diagrams.
This article explains how OpenClaw functions as an AI Agent framework that connects chat applications to large language models, manages multi‑turn dialogues, executes tool commands, handles memory and security, and demonstrates advanced features such as sub‑agents, cron jobs, and context compression.
The article explains AI agents as autonomous systems that perceive environments, plan, use tools, iterate through action loops, and self‑reflect, contrasting them with traditional chatbots and workflows, and outlines their core abilities, memory types, tool‑use mechanisms, and single‑ versus multi‑agent architectures.
This article explains why large language models need tool use, defines the concepts of Tool Use, Tool Call, and Function Calling, compares them, walks through a complete tool‑use workflow, and discusses architectural, safety, and design considerations for building reliable LLM agents.
This article explains why large language models alone can hallucinate, introduces the tool‑using agent architecture, and provides a step‑by‑step Python tutorial using LangChain, LangGraph, and Tavily to create, run, and evaluate a real‑time web‑search capable AI assistant.
This comprehensive guide walks you through the Claude Certified Architect exam structure, explains how Claude Code turns a language model into a full‑featured programming assistant, details installation on macOS, Linux and Windows, demonstrates tool‑use mechanics, context management, custom commands, and extending functionality with MCP servers such as Playwright.
Recent studies reveal that in Agentic RL, jointly training reasoning and tool-use on shared parameters creates a persistent negative interaction, with gradients nearly orthogonal, limiting performance; a disentangled tuning approach (DART) using separate LoRA adapters isolates the two abilities and restores gains across benchmarks.
This article shows how to create an Easysearch intelligent assistant that lets users manage indexes, write data, search and aggregate documents using Chinese natural language, by combining the DeepSeek large‑language model with OpenAI‑compatible function calling (Tool Use) and a lightweight Node.js executor.
This article walks through converting the Easysearch command‑line assistant into a web‑based tool by adding an Express API layer, reusing shared logic, and building a lightweight HTML/CSS/JS front‑end, while preserving the original CLI capabilities.
This tutorial extends a minimal LLM Agent loop by introducing Tool Use (function calling) to give the agent actionable capabilities—searching the web, retrieving the current datetime, and performing mathematical calculations—while explaining the BaseTool architecture, registration process, system‑prompt adjustments, and practical execution examples.
This article details the design, challenges, and performance gains of an AI‑driven rental recommendation system that replaces a multi‑agent architecture with a single LLM using dynamic tool‑use, introduces a two‑stage reinforcement‑learning pipeline, and achieves sub‑second latency and higher accuracy for complex rental scenarios.
Alibaba's new AgeMem framework turns long‑term and short‑term memory management for large language model agents into a learnable reinforcement‑learning task, replacing handcrafted rules with a three‑stage training process and achieving significant benchmark gains.
This article explains the ReAct framework, which augments large language models with explicit reasoning, tool use, and observation loops to overcome hallucinations, improve transparency, and enable dynamic interaction with external environments across diverse tasks.
The article breaks down how AI product managers can transform a raw base model into a market‑ready, high‑impact product by applying supervised fine‑tuning, tool‑use routing, RLHF alignment, and chain‑of‑thought reasoning, while highlighting trade‑offs, cost shifts, and evaluation metrics.
This article explains the origins of AI Agent Skills, walks through the standard tool‑use loop, provides step‑by‑step code for defining and invoking Skills with the OpenAI API, compares Skills with the emerging Model Context Protocol (MCP), and offers practical guidance for developers and architects building AI‑enabled systems.
Claude Opus 4.5, Anthropic’s latest flagship LLM, dramatically upgrades reasoning, tool use, and multi‑step automation, targeting high‑intensity enterprise scenarios, offering stronger coding, longer context handling, and better cost‑effectiveness, while still requiring careful prompt engineering and budgeting for token usage.
This article explores how to create a self‑thinking AI data analyst by detailing agent fundamentals, core modules such as planning, memory and tool scheduling, practical development steps, multi‑agent collaboration, evaluation benchmarks, and real‑world examples like stock backtesting.
The article evaluates nine cutting‑edge AI models on 150 simulated workplace tasks, showing that even the strongest models complete fewer than 40% of tasks, and uses these results to propose a hierarchical framework of agentic capabilities ranging from tool use to common‑sense reasoning.
VitaBench, a new open‑source benchmark from Meituan’s LongCat team, evaluates LLM‑driven agents across three realistic life‑service scenarios—food ordering, restaurant dining, and travel planning—using 66 tools and quantifying reasoning, tool, and interaction complexities, exposing a mere 30% success rate on complex cross‑scene tasks.
AgentFlow, a Stanford-led multi‑agent system built on a 7B model, outperforms massive models like GPT‑4o across ten benchmarks by leveraging modular agents, on‑policy learning, and a novel Flow‑GRPO training engine that solves sparse‑reward, long‑horizon challenges.
VitaBench is a newly released, highly realistic benchmark that evaluates large‑language‑model agents across three everyday scenarios—food ordering, restaurant dining, and travel planning—by quantifying reasoning, tool‑use, and interaction complexities, revealing a significant performance gap in current models.
This article analyzes the rStar2‑Agent paper, revealing how Agentic Reinforcement Learning, the GRPO‑RoC algorithm, a high‑throughput code‑execution service, and a three‑stage training recipe let a modest 14‑billion‑parameter model surpass much larger LLMs on challenging math benchmarks.
This article explains AI Agents by contrasting them with traditional AI tools, detailing their official definition, core components—planning, tool use, memory, action—illustrating a travel‑planning example, outlining agent types, and highlighting their significance for AGI and real‑world applications.
LongCat-Flash-Thinking, the latest open‑source model from Meituan, introduces domain‑parallel RL training, a high‑throughput DORA infra, and a dual‑path inference framework that together achieve state‑of‑the‑art performance on logical, mathematical, coding, and agentic tasks while maintaining top‑tier speed.
Meituan’s LongCat team unveiled LongCat-Flash-Thinking, an open‑source large language model that combines deep logical reasoning with tool‑calling capabilities, achieving state‑of‑the‑art performance across logic, mathematics, code, and agentic tasks, and introducing novel training frameworks such as domain‑parallel RL and DORA.
This article explains why the Model Context Protocol (MCP) is needed for LLMs, describes its client‑server architecture, data and transport layers, and provides step‑by‑step Python examples for creating both an MCP server and a client using FastMCP and low‑level APIs.
Kimi K2, an open‑source trillion‑parameter AI model released by Moonshot AI, offers Base and Instruct variants, achieves leading scores on benchmarks such as SWE‑bench, LiveCodeBench and AceBench, and introduces a novel post‑training autonomous‑exploration stage with MuonClip optimization to enable robust tool use and reinforcement‑learning‑driven self‑improvement.
AI Agents, autonomous intelligent programs that perceive, reason, and act, are reshaping industries from healthcare to autonomous driving; this article explains their core components, differences from large language models, planning techniques, memory mechanisms, tool use, real‑world applications, current challenges, and future directions.
This comprehensive guide walks you through the fundamentals of Retrieval‑Augmented Generation (RAG) and AI agents, explains their inner workings, shares optimization tricks, provides ready‑to‑run code snippets, and demonstrates how to evaluate performance with metrics such as recall, faithfulness, and answer relevance.
This article explains the importance of planning for multi‑tool AI agents, compares OpenAI and Anthropic approaches, presents experimental results, and provides practical guidance on tool design, prompt configuration, model selection, and parallel versus serial tool calls to improve efficiency and effectiveness.
This article demystifies AI agents and agentic workflows, explaining their core components—LLMs, tools, and memory—while detailing planning, tool‑use, and reflection patterns, comparing agentic, non‑agentic, and traditional workflows, and exploring real‑world applications, advantages, and limitations.
AI agents are rapidly evolving toward a 2025 breakthrough in perception, autonomous planning, tool use and memory, driven by multimodal models, neural‑symbolic reasoning and embodied intelligence, with $27 billion investment forecasts, exemplified by general‑purpose agents like Manus and emerging applications in code generation, research, healthcare, and risk analysis.
Professor Li Hongyi’s lecture provides a comprehensive, step‑by‑step exploration of AI agents, covering their definitions, reinforcement‑learning roots, LLM integration, memory mechanisms, tool usage, planning strategies, benchmarks, and practical examples, offering a valuable resource for anyone studying modern artificial intelligence.
This article breaks down the key endpoint features of large language models—such as quantization, max token limits, streaming cancellation, tool support, and reasoning ability—explaining what each term means, why it matters, and how to choose models wisely for different applications.
This article provides a comprehensive overview of AI Agents, tracing their evolution from traditional chatbots to LLM‑driven agents, explaining core components such as perception, reasoning, action, knowledge bases, learning and communication interfaces, and discussing practical use cases, interaction cycles, and future prospects.
The article explains what AI Agents are, how they differ from single large‑model systems, outlines four agent architectures—Reflection, Tool Use, Planning, and Multi‑Agent—and cites concrete examples from Grammarly, VS Code plugins, Image Describer X, ChatDev, as well as initiatives by Tencent and Google, highlighting the 2025 AI Agent boom.
Andrew Ng highlights that large language models are shifting from optimizing simple question‑answering for consumers to supporting complex agentic workflows, including tool usage, computer interaction, and multi‑agent collaboration, signaling a major evolution in AI capabilities.
This article surveys the rapid rise of AI agents powered by large language models, explaining their core perception‑planning‑action loop, memory architectures, tool‑use mechanisms, self‑reflection techniques, and real‑world case studies while highlighting current challenges and future prospects for autonomous intelligent systems.
AI agents extend large language models with planning, short‑term and long‑term memory, and tool‑use capabilities, enabling autonomous task decomposition, external API interaction, and persistent knowledge retrieval; frameworks such as MetaGPT, LangChain, and CrewAI simplify building agents like a researcher that gather information, browse web content, and generate reports, heralding broader AI‑enhanced productivity.
The article explains Andrew Ng’s four AI agent design patterns—Reflection, Tool Use, Planning, and Multi‑agent Collaboration—illustrates each with concrete Coze workflows and real‑world examples such as automotive research briefs, milk‑tea industry reports, and high‑quality travel planning.
The article dissects the rise of AI agents—from OpenAI's Assistants API and multimodal perception‑brain‑action pipelines to retrieval‑augmented generation, tool‑use strategies, single‑ and multi‑agent deployments, and emerging frameworks like AutoGen—while highlighting concrete examples, benchmark results, and current limitations.
The survey shows that once language models exceed roughly ten billion parameters they spontaneously acquire two complementary abilities—step‑by‑step reasoning, often elicited by chain‑of‑thought prompts or scratch‑pad training, and the capacity to invoke external tools such as search engines, calculators, or robots—enabling them to retrieve up‑to‑date information, perform complex computations, and act in the world, thereby advancing toward general artificial intelligence.
