Agents & MCP

Reported

Agent.reviews launches as a platform where AI agents read and write tool reviews

Louis, co-founder of Armature (YC P26), launched Agent.reviews, a site where AI agents can read and write reviews on developer tools based on observed usage patterns.

Reported

Andrej Karpathy highlights the extreme stratification of AI usage, from casual LLM users to a tiny elite with internal frontier access

A Hacker News post by Andrej Karpathy breaks down the AI user base into tiers, revealing that only about 5,000 people globally have access to frontier systems internally.

Primary source

Apify blog compares AI agents and MCP servers, advising which to build

Apify published a blog post explaining the differences between AI agents and MCP servers, covering ownership, context costs, and build decisions.

Primary source

Apify blog details MCP connector delivery for pay-per-result Actors

Apify published a build log showing how to add MCP connector delivery to a pay-per-result Actor, covering input schema, argument mapping, and live delivery into Notion.

Primary source

Apify blog post explains why many Actors are invisible to AI agents and how the MCP server fixes them

An Apify blog post reveals that popular Actors like a LinkedIn Sales Navigator scraper are unusable by AI agents until the MCP server transforms their inputs and outputs.

Primary source

Apify blog post reveals AI agents silently failing when calling Actors

A developer at Apify found that 5 out of 16 Actors returned empty results to AI agents while reporting success, prompting fixes to improve agent reliability.

Primary source

Apify blog shows how to build a Grok Bot that tracks viral posts across X, TikTok, and Instagram

Apify published a tutorial on building a Grok Bot that scrapes social media platforms overnight, ranks posts by save rate, and sends the top three to the user each morning.

Primary source

Apify blog shows how vague input schemas trip up AI agents using its Reddit Actor

A developer describes connecting their Apify Reddit Actor to an AI agent via the Apify MCP server, only to find the agent produced garbage data because the input schema lacked clear descriptions.

Primary source

Apify compares its RAG Web Browser with Jina AI Search for LLM evidence retrieval

Apify published a benchmark comparing its RAG Web Browser to Jina AI Search on a single LLM briefing task, evaluating usable evidence retrieved and cost.

Primary source

Apify compares Parallel's Extract API with its own Web Fetch for agent-based web reading

Apify published a blog post comparing Parallel's Extract API to its own Web Fetch tool, focusing on the trade-offs between search-oriented retrieval and direct extraction.

Primary source

Apify integrates with LangGraph for real-time web data access

Apify announced an integration enabling LangGraph agents to fetch real-time web data via its official langchain-apify package or Apify MCP Server.

Primary source

Apify lists 12 LangChain alternatives for AI agent development in 2026

Apify published a blog post comparing 12 frameworks that can replace LangChain for building AI agents.

Primary source

Apify migrates MCP server to new stateless specification

Apify has migrated its MCP server to the July 2026 stateless spec revision, detailing the changes and implications for hosted servers and existing clients.

Primary source

Apify publishes beginner's guide to Model Context Protocol

Apify released a tutorial explaining what the Model Context Protocol is and how to connect an AI agent to its MCP server.

Primary source

Apify publishes tutorial on using MCP to connect AI agents to Actors

Apify released a guide showing how to use the Model Context Protocol (MCP) to give AI agents like Claude access to live web data via Apify Actors.

Primary source

Apify shows how to build an AI research assistant in Slack using its MCP server and n8n

Apify published a tutorial on integrating its MCP server with n8n to create a Slack-based AI research analyst that answers team questions with live web data.

Primary source

Apify tests Claude with and without its MCP server for real-time product data

Apify published a side-by-side comparison of Claude using its Model Context Protocol server versus relying on the model's built-in knowledge for product data retrieval.

Primary source

Bright Data Blog covers Microsoft Agent Framework with Web MCP

Bright Data published an article exploring the integration of Microsoft's Agent Framework with the Web MCP (Model Context Protocol).

Primary source

Bright Data explains Context as a Service for AI data retrieval

Bright Data published a blog post defining Context as a Service, a concept where external context is provided to AI models via structured data feeds.

Primary source

Bright Data integrates Chrome DevTools with MCP protocol

Bright Data published a blog post detailing how to use Chrome DevTools with the Model Context Protocol (MCP).

Primary source

Bright Data publishes guide on lightweight SERP grounding using its MCP server

Bright Data's blog describes a method for grounding AI agent outputs with real-time search engine results page data via its Model Context Protocol server.

Primary source

Browserless introduces Skill Bucket for agent-based browser automation

Browserless announces a new Skill Bucket feature that allows AI agents to access and use predefined browser automation skills.

Primary source

Browserless launches Agent for AI-driven browser automation

Browserless has introduced a new product called Browserless Agent, designed to enable AI agents to control browser sessions.

Primary source

Browserless publishes guide on E2E testing with Claude Browser Agent

Browserless published a blog post about using Claude Browser Agent for end-to-end testing.

Primary source

Browserless releases Hermes Agent Plugin for AI agent integration

Browserless has published a plugin that connects its browser automation service to the Hermes AI agent framework.

Primary source

Claude invents Google Trends rankings after refusing to compare across batches

A blog post on Apify describes how Claude refused to rank Google Trends data across separate batches due to API limitations, then fabricated a ranking anyway.

Primary source

Crawlbase compares nine MCP servers for web scraping

Crawlbase published a blog post evaluating nine MCP servers for web scraping, covering tools, hosting, anti-bot handling, and pricing, with a setup guide for its own Crawlbase Web MCP in Claude.

Primary source

Developer wraps Spanish public-tenders scraper as MCP tool for Claude Desktop

A developer describes how they turned an Apify Actor that scrapes Spanish public tenders into a callable tool for an AI agent via the Apify MCP server.

Primary source

Firecrawl compares OpenAI Decisions API and JEV

Firecrawl published a blog post comparing the OpenAI Decisions API with JEV.

Reported

OpenAI uses thousands of AI agents to analyze complex biological data

OpenAI published a post describing how it deployed thousands of AI agents to uncover new leads in biological data.

Primary source

Playwright v1.64.0 adds WebMCP support for testing browser-based MCP tools

Microsoft released Playwright v1.64.0, introducing `page.webmcp` and `frame.webmcp` methods to interact with tools registered via the experimental WebMCP browser API.

Primary source

Scrapfly compares five MCP servers for web scraping and browser automation

Scrapfly published a blog post comparing five MCP servers by their capabilities in protected-site scraping, browser control, debugging, static fetching, and cross-browser automation.

Primary source

Scrapfly ranks 7 AI browser agents for production scraping in 2026

Scrapfly published a ranked guide to the best AI browser agents for automation and scraping, evaluating them on production stability and anti-blocking capability rather than demo performance.

Primary source

ScrapingBee explains how CodeWhale gives AI agents live web access

ScrapingBee publishes an article detailing how CodeWhale enables AI agents to access live web data by integrating web scraping tools, MCP servers, or framework tools.

Primary source

ScrapingBee explores how 'Agent Skills' can make AI coding agents more reliable

ScrapingBee published a blog post examining the Agent Skills repository by addyosmani, which aims to improve the reliability of AI coding agents by teaching them structured workflows.

Primary source

ScrapingBee publishes blog post on /last30days AI agent skill for multi-platform research

ScrapingBee's blog introduces /last30days, an AI agent skill designed to research topics across multiple internet platforms while navigating access restrictions.

Primary source

ScrapingBee publishes guide on MCP servers for web scraping, emphasizing control over data

ScrapingBee explains how a Model Context Protocol (MCP) server can improve scraping agent performance by keeping context lean and avoiding page bloat.

Primary source

ScrapingBee publishes guide to AI agent web scraping with wigolo and MCP

ScrapingBee released a guide covering how to equip AI agents with web scraping capabilities using wigolo and the Model Context Protocol.

Primary source

ScrapingBee publishes tutorial on building a live data retriever with Dify AI agents

ScrapingBee's blog shows how to use its Remote MCP server as a live data retriever tool within Dify AI's open-source agentic workflows.

Primary source

ScrapingBee publishes tutorial on building an AutoGPT agent with its API

ScrapingBee released a blog post showing how to use its web scraping API as a data source within AutoGPT's visual Agent Builder.

Primary source

Scrapingdog launches MCP Server for AI-powered web scraping

Scrapingdog has introduced an MCP Server that integrates its web scraping and data extraction capabilities directly into AI-powered applications.

Primary source

SerpApi adds MCP Bundle support to its MCP server

SerpApi has released MCP Bundle (MCPB) support, allowing users to download the server as a .mcpb file for installation in compatible desktop applications.

Primary source

SerpApi launches OpenCode plugin for real-time web search in AI coding agents

SerpApi released an open-source plugin that integrates its search API into OpenCode, enabling AI coding agents to query Google, Amazon, YouTube, and other sources in real time.

Primary source

SerpApi publishes guide on integrating Hermes Agent with real-time web search

SerpApi released a tutorial explaining how to use Hermes Agent for real-time web search, monitoring, and research workflows.

Primary source

Stagehand adds Claude toolset integration guide to v4 docs

Stagehand published a documentation guide for connecting Claude to local Chrome or Browserbase via its Claude SDK packages.

Primary source

Stagehand Python adds WebMCP tool support inside iframes

Stagehand Python's latest dev release enables discovery and invocation of WebMCP tools registered inside iframes by routing calls through the main CDP session and all adopted OOPIF sessions.

Primary source

Zenrows integrates with smolagents to give AI agents production-grade web access

A tutorial shows how to swap smolagents' plain-requests VisitWebpageTool for a Zenrows-backed tool that bypasses bot checks.

Primary source

Zenrows MCP brings live scraping to Cursor's AI editor

Zenrows released an MCP integration that lets Cursor users scrape JavaScript-rendered and bot-protected sites directly from the editor.

Primary source

Zenrows publishes tutorial on building a web-aware TypeScript agent with Mastra

Zenrows published a blog post showing how to combine its Fetch tool with the Mastra agent framework to give AI agents reliable access to dynamic and protected web pages.

Primary source

Zenrows publishes tutorial on building a web research multi-agent system with AG2

Zenrows released a blog post showing how to integrate its Fetch tool as a typed tool in an AG2 multi-agent group chat to ensure clean content circulates among agents.

Primary source

Zenrows publishes tutorial on building an AI lead generation agent with OpenAI Agents SDK

Zenrows released a blog post demonstrating how to combine its scraping API with OpenAI's Agents SDK to build a lead generation agent that reads bot-protected directories, enriches company data, and scores leads against an ideal customer profile.

Primary source

Zenrows shows how to build a web-aware agent by combining its fetch tool with OpenAI Agents SDK

A tutorial from Zenrows demonstrates registering the Zenrows Fetch tool as a function tool alongside OpenAI's WebSearchTool to let an agent choose between discovery and full-page retrieval.

Primary source

Zyte CDP enables agentic browser automation via natural language instructions

Zyte published a blog post demonstrating how its CDP (Chrome DevTools Protocol) service can be used to let AI agents control a real browser through plain-language commands.

Primary source

Zyte demos self-healing Scrapy spiders that repair crawls after site redesigns

Zyte published a blog post detailing two workshop demos from Extract Summit Austin where an agent fixes a Scrapy spider after a site redesign and heals a live crawl mid-run using Spidermon and the Scrapy MCP, including cost measurements and design mistakes.

Primary source