#AI Agent
1088 related articles
ToDo.is: Hand Your To-Dos to an AI Agent That Actually Does the Work
ToDo.is is an always-on AI agent to-do tool: write a task and it executes—research with sources, generate files and images, scheduled briefings and price alerts. Works on web, Android, WhatsApp, and Telegram, and asks before paid, send, or delete actions.
MCP Protocol Deep Dive: A Standardized Connection Solution for AI Agents
A deep dive into the Model Context Protocol (MCP): its Host/Client/Server architecture, JSON-RPC messaging, STDIO and Streamable HTTP transports, Tools/Resources/Prompts primitives, and security best practices—helping AI developers systematically understand this standardized agent connection protocol.
Give Your AI Agent a Shelf: MCP + x402 Lets Agents Shop Autonomously
A real experiment: give an AI Agent a secondhand shelf, expose search, lookup, and purchase-intent tools via MCP, and return payment challenges with x402—but eBay still closes the deal. Exploring agent-readable e-commerce.
Open-Source Tool StepFork: Record and Replay AI Agent Executions So Debugging No Longer Relies on Luck
StepFork is an open-source AI Agent debugging tool that supports recording and replaying execution traces, forking experiments to fix issues, minimizing failed traces, and generating regression tests. v0.2.1 is seeking contributors for cross-platform CI, CLI testing, and LangGraph tutorials.
AI Agent Full-Stack Development Course: LangChain + LangGraph + MCP in Practice
A practical breakdown of the Bilibili AI Agent full-stack course: Vue project demos explain the perceive-decide-act loop, plus LangChain, LangGraph, MCP, and LLM selection.
WorkBuddy Hands-On Tutorial: Getting Started with Desktop AI Agent Office Automation
WorkBuddy desktop AI Agent hands-on tutorial: installation, claiming credits, file organization, image-to-video, personal website development, Feishu connectors, and automated scheduled tasks—covering three working modes and the three ecosystems of skills, experts, and connectors, with a pitfall-avoidance guide.
WorkBuddy from Beginner to Expert: How AI Agents Actually Operate Your Computer
A hands-on review of Bilibili creator Shanyao's WorkBuddy tutorial, analyzing how AI Agents upgrade from ordinary large models that "give solutions" to ones that "deliver results," batch-converting Word to PDF with automatic archiving, plus credit consumption and use-case analysis.
Getting Started with AI Agents: Core Definitions and AI Agent Workflow Explained
Explaining the core definition of AI Agents: LLMs combined with tools, completing tasks through reasoning, tool calling, and iterative loops. Comparing AIGC and AI Agents across core capabilities, interaction models, outputs, and external interaction, and exploring the 2026 Agent surge and development practices.
Browser AI Agent Comparison: How to Choose Among Browy, Browser Use, Skyvern, and Five Other Tools
A side-by-side comparison of seven browser AI agents—Browy, Browser Use, Skyvern, Nanobrowser, Stagehand, Claude in Chrome, and ChatGPT—covering where they run, whose login session they use, who pays for the model, and which use cases each fits best.
Laminar: Using AI Models to Catch Silent Agent Failures
Laminar (YC S24) launches on Hacker News with a purpose-trained model to detect silent failures in AI Agents—semantic errors that logs and metrics miss. We analyze the approach, its value, and open questions.
AI Agents Blocked by Airlines, Banks, and Retailers: The Web Is Building Walls
Yelp, Delta, JPMorgan Chase, and other major services are blocking AI agents, with Cloudflare's default changes widening the blockade. We break down deliberate corporate blocking versus unintended security collateral damage, the latest in Amazon v. Perplexity, and whether Meta-led open standards can reopen the door for AI agents.
Building an Agent on AWS: Score GitHub Repos Before You Clone
FakeGit involved 7,600 malicious GitHub repos and 14M downloads, and agents recommend them. RepoPilot uses AWS Lambda, Nova Pro, and Strands to give any repo a 0-100 trust score and flag dangerous install steps before you clone.
Amazon AGI Lab Open-Sources Perception Agents: Teaching AI to Read the Screen
Amazon AGI Lab open-sources visual annotation and verification perception agent primitives at AI Engineer, letting AI understand screen content and closing the loop from design feedback to code merge.
WorkBuddy Beginner's Guide: How AI Agents Take Large Models from "Talking" to "Doing"
Based on a Bilibili creator's hands-on demo, this guide explains how WorkBuddy AI Agent handles batch PDF conversion and file archiving, contrasts chat-style models like Doubao, and breaks down the full Agent workflow from understanding needs to delivering results, with credit mechanics and beginner tips.
Cursor Engineering Lead: How to Give AI Agents Their Own Cloud Computers
Cursor's engineering lead explains how to give AI agents cloud computers—from cloud onboarding agents and the AnyDev CLI to Agent Experience and self-improvement, covering the three-stage agent engineering path and computer use as the next key frontier.
MCP Server Post-Mortem: 8 Lessons from Redesigning My Internal Tools
An MCP server post-mortem: from generic run_sql to narrow tools, user identity propagation, docstrings as interfaces, structured errors, resource data leakage protection, server-side auditing, and server versioning—8 hands-on lessons with Python code examples.
Hands-On Test: Claude Opus 5.5, GPT-6 Astra, and Gemini 4 Argon — Three Projects Settle the Score
Hands-on test of Claude Opus 5.5, GPT-6 Astra, and Gemini 4 Argon on three real coding projects: a 3D game, Blender modeling, and a Notion clone. Benchmark scores were close, but real gaps widened: Opus was highest quality, Astra the best value, and Argon not yet in the same tier.
The Battle for AI Agent Control: Who Owns Identity, Rules, and Budget?
Google, Workday, and OpenAI drew AI Agent boundaries within a week. We analyze the fight over Agent identity, permissions, and budget, with MCP, CERT-In advisories, and the European supermarket attack, plus an enterprise checklist and lab test methods.
When AI Agents Are Left Unguarded: 6 Out of 10 Make Themselves Admin?
A Kaggle Benchmarking Challenge experiment had 10 AI Agents perform tasks in an open-permission environment; 6 made themselves admin. This article analyzes causes, evaluation methodology, and engineering defense.
RAG Retrieval Gone Wrong? The Real Problem Lurks in Chunk Boundaries
RAG returns irrelevant results? The problem is often chunk boundaries and metadata design, not embedding models. A practical look at Agent overuse, framework wars, RAG chunking, and observability.