"Information on agent swarm architecture" matching MCP servers:

Enables browser automation with concurrent tab pool management using Puppeteer. Supports navigation, content extraction, screenshots, element interaction, and JavaScript execution across multiple browser tabs with auto-recovery and idle timeout features.

Web agent that keeps working when a task leaves the browser: deterministic routing between the page DOM and Windows UI Automation, with LLM-free skill replay.

Enables MCP clients to drive a self-hosted Chrome session for browsing tasks through browser tools, while providing a live view URL so a human can take over at prompts like logins or 2FA.

Enables an agent to run whole browser tasks on live web pages from a plain-language goal — opening tabs, clicking, typing, scrolling, and returning the relevant page sections as markdown — with every step reported alongside its confidence. Tabs can be reused and closed across calls, so multi-step flows continue on the page the previous task ended on.

Enables AI agents to automate a Linux desktop natively on Wayland, with 68 tools for screen capture, input simulation, browser automation, OCR, accessibility, media, audio, clipboard, notifications, filesystem, system, memory, and network operations through the MCP protocol.

A self-hosted MCP server that collects links and articles from web and mobile, allowing AI agents like Claude or ChatGPT to read and analyze them as a reading list.

Enables AI agents to control a real, persistent Chrome browser over MCP, perceiving pages as accessibility trees and interacting with elements by their accessible names without automation fingerprints or site-specific glue.

Provides a headless Chromium browser through MCP, enabling AI agents to browse JavaScript-rendered pages, search the web, capture screenshots, extract tables and data, and run stateful multi-step interactions like clicking, typing, and form submission.

MCP server that wraps Playwright to give AI agents eyes on the web, enabling browser search, navigation, extraction, and interaction with intelligent LLM-based DOM extraction and skill caching.

A multi-function Streamable HTTP MCP tool aggregation server that provides web search via Brave, Exa, and SearXNG with multi-key rotation and cross-provider fallback, and supports extensible tool families (URL fetch, code search, RAG) through a pluggable architecture.

Enables agent-native web search and multi-angle research synthesis with pay-per-call USDC payments on Base via x402, requiring no API keys or subscriptions.

Enables a person to point their own AI agent at specific on-screen app elements over MCP, letting the agent read the underlying data without sharing the screen or receiving visible content.

Zero-knowledge credential injection for AI agents. Your agent authenticates to websites and APIs without ever seeing a password, TOTP code, or API key.

Converts URLs into clean, LLM-ready markdown, respecting robots.txt and never bypassing anti-bot measures or paywalls.

Enables AI agents to scan any website for agent-readiness and generate the necessary artifacts (llms.txt, WebMCP scaffold, structured data) to make it agent-ready.

MCP server for intelligent web crawling using a ReAct agent. Adaptively explores websites and returns structured analysis.

Agent Bridge is an MCP server that enables local agents like Codex and Claude Code to remotely control a logged-in ChatGPT Pro session in the browser, supporting automated discussions, reviews, and code implementation with secure pairing and multi-session coordination.

A local Streamable HTTP MCP server that lets several AI agents drive your real, logged-in Chrome at once, each isolated in its own background tab group without stealing focus. It exposes browser tools for navigating, reading pages, filling forms, signing in via Bitwarden, and handing steps back to the user.

Lets MCP-speaking terminal agents read, click, type, fill forms, upload files and take screenshots in the user's own already-logged-in Chrome, so existing sessions stay signed in. Connects to the current browser via a Chrome extension instead of launching a fresh, empty one.

Free multi-source search for AI agents with multi-source verification, token savings, and MCP native support.

Enables MCP-compatible agents to perform web searches via the agent-web-search engine, returning ranked results with sources.

MCP server that gives AI agents browser-based access to Mobbin for design research, screen discovery, and screenshot collection through automated browser control.

Enables AI agents to access authenticated websites by reusing your existing browser profile and cookies, so you never share passwords and sessions persist between runs.

Enables AI agents and MCP clients to drive real Chrome, Edge, Brave, and Firefox browsers through native automation protocols. It exposes tools for navigation, page snapshots, form filling, clicks, screenshots, console/network inspection, and tab management.

Provides six general-purpose utilities (geocoding, reverse geocoding, weather, web search, URL-to-Markdown, secure random) as paid MCP tools, with each call costing $0.01 USDC on Base mainnet.

Sub-15ms edge intelligence tools for Claude Desktop, Cursor, and LLM agent loops. Provides sub-millisecond LLM JSON repair, web-to-markdown token compression (cuts context tokens up to 90%), PII redaction, and RFC 5322 B2B email deliverability verification at zero marginal token cost.

The MCP for the Web Speed Agent SDK that enables post-auth agents.

Paid web research MCP tools for autonomous agents: search, page extraction, citations, and diff monitoring through a live x402 API. Unpaid calls return the Base USDC payment requirement so agents can pay and retry safely.

This server allows users to search, read timed subtitles, and sample frames from Bilibili videos. It enables the automated saving of recommended content to a user's watch-later list or favorite folders while maintaining local session security.

Enables Claude to delegate web search, page reading, text drafting, and parallel multi-model image generation to cheaper helper models on any OpenAI-compatible backend. Claude stays in charge of planning and final decisions while consuming fewer of its own tokens.