nvda-mcp server
Related Servers
Alternatives to nvda-mcp server
No user-submitted related servers found.
Related Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to inspect and automate Windows desktop UI elements by exploring UI trees, checking properties, performing actions like clicking and typing, and generating Python automation scripts.18 PyPIMIT
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to query a local reader's current state (book, chapter, CFI, visible text, selection) and perform actions like opening books, navigating by CFI or progress, searching, and creating or clearing highlights.MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to externally drive a running ArcGIS Pro instance to run GP tools and Python, with real-time human-readable feedback in the Pro panel.7MIT
- AlicenseNot gradedqualityBmaintenanceEnables an AI to see and drive any Windows application through the accessibility tree - reading real element names, states and actions and invoking the app's own controls without screenshots or moving the mouse, even for windows behind others. Also exposes 200+ native Windows tools for system health, hardware, network, privacy, files, processes, registry, media and more.3Apache 2.0
- FlicenseNot gradedqualityCmaintenanceEnables AI assistants to control Windows GUI by listing and focusing windows, capturing element snapshots via UIA/OCR/CDP, performing clicks/inputs/scrolls, verifying changes, waiting for screen updates, taking screenshots, and obtaining visual descriptions.2-
- AlicenseNot gradedqualityFmaintenanceEnables AI assistants to interact with VS Code for language intelligence, debugging, and code execution.17MIT
TDQS
Scored across 37 tools
Most tools target clearly distinct operations, and the descriptions actively steer between near neighbours (get_object_tree vs find_objects vs get_text, type_text vs set_value vs press_keys, get_log vs get_action_log vs get_speech_history). A few pairs still blur: get_focus vs get_navigator_object, and inspect_object vs get_dev_info both cover API/class information for a control, so an agent may hesitate between them.
Every tool carries the nvda_ prefix followed by a snake_case verb_noun pattern (get_focus, find_objects, press_keys, activate_window, list_scripts, run_scenario). The only bare verb is nvda_speak, which reads naturally and does not break the scheme.
37 tools is heavy and well past the comfortable range, but the server legitimately spans several sub-domains: perception, input control, add-on development, and accessibility regression testing. Each tool earns a place, yet the surface is large enough that discovery and selection cost is real.
Coverage is unusually thorough: perception (tree, text, focus, speech, output history), control (keys, typing, activate, focus, set value), diagnostics (logs, event tracing), add-on development (API lookup, HTML inspection, dev info) and testing (focus-order audit, scenarios, speech preview). Minor gaps remain, such as reading or setting configuration rather than only reading it, and capturing a full-window screenshot rather than a single control.