1495 Model Context Protocol servers — searchable, rated, ready to install.
Chrome DevTools for AI assistants: console, network, DOM, source maps, breakpoints, live variables.
Paste a TripAdvisor hotel, restaurant or attraction URL: reviews are collected for you and turned…
Crawl a site for broken links, redirects and missing SEO tags. Pay per page on Apify.
Full-page website screenshots and visual diffs between two captures. Pay per capture on Apify.
Records your screen and voice, returns one annotated screenshot and finding per mark.
{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"whoami","arguments":{}}} free no key
Scrape SmartRecruiters company job boards and get clean, de-duplicated JSON: title, department…
Deterministic webpage change detection, hashes and text diffs for recurring monitoring via x402 MCP.
Playwright-style MCP to inspect, control, and debug PySide6 desktop GUIs.
Annotated-screenshot reviews and ready-made product studies your AI agent can read, run, and share.
Token-first headless browser for AIs. Extracts a tiny semantic graph, cutting HTML bloat by 95%.
Extract public webpages as Markdown, map sites, and run bounded crawl jobs. API key required.
Apps, games and extensions with ratings and user reviews from app stores, Steam and Chrome.
30+ web-access and AI APIs behind one key: search, scraping, browsers, voice, OCR and LLMs.
Find Reddit leads and brand mentions. AI buyer labels and Slack or Notion delivery. $1.50/1k results
44 tools for AI agents: Trends, SERP, SEO audits, screenshots, weather, FX, whois, RSS.
Use Reddit, Zhihu, X and WhatsApp through your signed-in browser with compact, on-demand MCP tools.
Telegram/VK audience scraping, marketplaces, Russian government registers. Free open dataset.
Replayable browser automation: Jev picks the element, a gate blocks risky actions, every run traces.
macOS automation: 17 typed tools for windows, menus, keyboard, clipboard, tabs, screenshots
Google Maps scraper for real estate agencies and agents in any city.
Check what AI crawlers may read and what a page permits, generate llms.txt and schema.
Web scraping runtime for agents: parsers, scrapes, batches, crons, and SQL databases.
Checks robots.txt line by line and says what each AI crawler token actually controls - training, AI
POST /mcp