"Slow thinking, distributed thinking, and reasoning abilities research" matching MCP servers:

Windows桌面截图与窗口自动化工具，支持后台操作、不抢焦点，可截取窗口/屏幕、模拟鼠标键盘、点击菜单等。

A specialized MCP server that helps analyze and debug Model Context Protocol logs by providing Claude with direct access to log files across multiple platforms.

Enables local computer diagnostics by sampling CPU, memory, disk, and network activity and displaying the busiest processes in an interactive dashboard. It also supports human-confirmed process-stop proposals while keeping diagnostics local and bounded.

Enables AI agents to plan, confirm, submit, observe, and audit compute jobs across local machines, SSH workstations (including WSL), and Slurm clusters through a unified stdio interface, with node probing, environment evidence, project sync, storage governance, and progress/ETA tracking.

Enables cluster-aware command execution and automatic task routing across distributed nodes based on system load, architecture, and OS requirements. It supports parallel execution, remote node management via SSH, and dynamic load balancing for agentic workflows.

Enables AI agents to capture and inspect a window, send background clicks, keys, text and scrolls, and write, validate, run and monitor YAML automation scripts (or watchers) so fast reactive work is handled without the agent's thinking latency.

An MCP and HTTP server wrapper for PyAutoGUI that enables LLMs to control mouse and keyboard operations across local or remote distributed environments. It provides comprehensive tools for screen interaction, including clicking, typing, and capturing screenshots.

Enables Codex sessions to observe and control the Windows desktop through screenshot reasoning, archived-frame inspection, pixel sampling, keyboard/mouse input, and workspace-local multimodal memory.

Lets ChatGPT Web act as the reasoning layer for an agent that runs shell commands, reads and writes files inside a constrained workspace, captures screenshots as image content, and controls the mouse and keyboard. It also keeps persistent Markdown memory and identity/skill context files that load automatically into the conversation.

Lets an AI agent work on a real Commodore Amiga over the home network: AmigaDOS Shell, files, ARexx, screenshots, mouse and keyboard, and windows, gadgets and menus by name. Includes a remote desktop to watch and take over, tuned for 68000 machines and slow links.

On-server agent for managing PostgreSQL, Redis, Keycloak, and NGINX on Ubuntu infrastructure with real-time monitoring, automated maintenance, and distributed command execution.

Lets MCP clients discover and run AppleScript or JavaScript for Automation (JXA) to control macOS applications, inspect the system, or reuse scripts from a bundled knowledge base.

MCP server for system monitoring, enabling users to ask about CPU, memory, disk, network, and processes on their system.

Wraps the mcp-locks CLI to expose coordination primitives as MCP tools, enabling multiple AI agents to claim, release, and list exclusive OS resources like browser profiles.

Lets AI assistants understand what you're working on — current screen content, recent dictation, clipboard, and saved notes — running entirely on your own machine with nothing sent to the cloud.

Allow AI agents to see and control a real Windows PC you own: observe (UIA + screenshots), click/type/drag/scroll, launch apps, owner Live View. BYOH — your machine, your key.

Executes privileged commands with mandatory human approval via a TUI prompt, supporting local and remote execution through sudo.

Enables LLM agents to monitor Mac thermals and control fan speed with safety limits enforced by a root daemon.

Let AI agents operate a real USB-connected iPhone from Linux — no Mac, no jailbreak, no paid developer account. Screenshots, taps, swipes, text input, element tree and app control via WebDriverAgent, with CoreDevice HID fallback.

A TypeScript-based server that enables programmatic management of pseudo-terminal sessions. It allows users to spawn, control, and interact with terminal processes in real-time through the Model Context Protocol.

Enables AI agents to inspect pending updates and service health for Proxmox VE VMs and containers, then take snapshots, apply official updates, reboot, and roll back under tag-based permissions and human-confirmed risky actions.

MCP server that reads and writes the system clipboard — tables, text, code, JSON, URLs, images, and more. Preserves spreadsheet structure (rows/columns) that is lost when pasting into Claude directly.

An MCP server for Hyprland that enables AI agents to control workspaces, windows, mouse, keyboard, and take screenshots on a Wayland desktop.

Enables reading, writing, and manipulation of Bike outlines on macOS through natural language, using AppleScript for local document management.

GenieACS-MCP is a Go-based MCP server that bridges any GenieACS (TR-069 ACS) instance, exposing device data, firmware management, and CPE actions (reboot, parameter refresh, firmware download) over JSON-RPC

Clones the OpenCode workbench installation onto Linux machines — locally or over SSH — reproducing the same OpenCode config, MCP servers, agent skills, plugins, and Docker tooling. Exposes 6 tools over stdio.

Enables MCP-speaking clients to control a real desktop via the computer_use tool, including clicking, typing, scrolling, dragging, key combos, app focus, and screen/accessibility capture. It also provides a verdict system that verifies whether input actions had their intended effect, with configurable approval modes.

MCP server for macOS disk cache management. Enables AI agents to free disk space on demand by scanning and clearing developer caches.

Control TraeCN desktop automation through explicit, profile-scoped MCP tools. Supports task delegation, model control, dialog handling, code review, and unattended workflows.

Enables AI assistants to control mobile and desktop devices with natural language, including running automation tasks, taking screenshots, and managing devices.