Инструменты

From Microsoft: converts PDF, Word, Excel, PowerPoint, HTML and other files or URLs to Markdown so agents can read and search them. A single convert_to_markdown tool over STDIO or local HTTP.

Gives agents long-term memory across sessions with a local knowledge graph of entities, relations and observations, stored in a local JSONL file you can query and edit. Official MCP reference implementation.

Lets agents drive a real browser through the accessibility tree: navigate, click, fill forms and take screenshots, without needing a vision model. Maintained by Microsoft.

Local MCP server built into Browser Use: lets agents navigate, click, type and extract page content, or hand multi-step tasks to the built-in browser agent. Bring your own OpenAI or Anthropic API key.

Official MCP reference server for reading, writing, editing, moving and searching local files within an allowlist of directories, adjustable at runtime through Roots. Handy for organizing documents from a desktop client.

Provides a sequential_thinking tool that guides the model to break complex problems into revisable, branching steps. Good for planning, analysis and tasks that need careful iteration. Official MCP reference server.

Official PaddleOCR MCP: text recognition and layout parsing for images and PDFs (PP-OCR, PP-StructureV3, PaddleOCR-VL), returning text or Markdown. Runs inference locally, or calls the official API or a self-hosted service.

MCP server built into the adaptive Python scraping framework Scrapling: supports plain HTTP requests, browser rendering and stealth fetching, and uses CSS selectors to pass only the target content to the AI, saving tokens and filtering out prompt injection.

Lets agents control and inspect a real Chrome: read the console and network requests, take screenshots, and record performance traces with optimization suggestions. Maintained by the Chrome DevTools team.

Browser automation MCP server open-sourced by ByteDance in the UI-TARS-desktop repo: drives pages through Puppeteer accessibility data with an optional vision mode, running locally over stdio or remotely over SSE or Streamable HTTP.

MCP built into the open-source social media scheduler Postiz: agents draft, schedule and publish posts and view analytics across 30+ platforms including X, LinkedIn, Instagram and YouTube. The cloud version uses browser OAuth, no API key (paid after a 7-day trial); self-hosting is free.

Trading research agent from HKU's HKUDS whose MCP server offers 70+ tools for quotes, financials, backtesting and research across A-shares, Hong Kong and US stocks and crypto. Needs local Python; HK, US and crypto research need no API key. For research only; not investment advice.

Runs the cognee memory engine as an MCP server whose remember, recall and forget tools write documents and conversations into a knowledge graph and vector store for cross-session retrieval. Start it with Docker or uv first; needs an LLM API key by default, or connect to Cognee Cloud.

Writes conversations and business data into Graphiti's temporal knowledge graph so agents can retrieve long-term memory by entity, fact and time. Open source from Zep; start FalkorDB or Neo4j and the server with Docker first, and configure an LLM API key.

Open-source SEO tool (a Semrush/Ahrefs alternative) with built-in MCP for keyword research, rank tracking, competitor and backlink analysis, site audits and AI search visibility checks. The hosted version has a free trial; self-hosting needs a pay-as-you-go DataForSEO key.

Continuously records your screen and audio locally, extracting on-screen text and transcribing speech, so agents can search pages, docs and meetings you saw and summarize your day. Data stays local. Needs screenpipe running: source is free for personal use, desktop app is a subscription.

Official MCP from the FunASR project: transcribes audio files locally with the SenseVoice model, supporting Mandarin, Cantonese, English, Japanese and Korean, with optional segment timestamps. Local inference needs no API key and audio stays on your machine. Docker image available.

MCP built into the CLI of the Maestro mobile UI testing framework (maestro mcp): lets agents view the screen, tap, scroll and assert on iOS/Android simulators, Android devices or Chromium, and write and run YAML end-to-end flows. Needs the Maestro CLI and Java 17+. Official.

A Chrome extension that hands the browser you use every day to your agent, keeping your logins and settings, with screenshots, network monitoring, bookmarks and history, and semantic search over tab content. Install the extension and mcp-chrome-bridge first.

Connects to the Arize Phoenix observability platform so agents can query LLM traces and sessions, manage prompts, browse datasets and read experiment evaluation results. Maintained by Arize; needs a self-hosted or cloud Phoenix instance.

Lets agents control iOS and Android simulators and real devices: reads the accessibility tree first and falls back to screenshot coordinates, and can swipe, install and uninstall apps, record the screen and read device logs. Requires Xcode or Android platform tools.

Lets agents operate the Windows desktop directly: launch apps, switch windows, simulate keyboard and mouse, read UI elements and run PowerShell, without needing a vision model. Windows only.

Deep research assistant that runs locally with a built-in MCP server: searches arXiv, PubMed, Wikipedia and more over multiple rounds and writes cited reports. Needs Python 3.12+, uv and an LLM (local Ollama or an API key); web search needs SearXNG or similar. Runs take minutes to half an hour.

Hands your current browser session to the agent through a Chrome extension: read console logs and network requests, take screenshots, and run Lighthouse performance, accessibility and SEO audits without signing in again.