dsh-plugin-computer-use
Run the following command in DeepSeek Harness:
dsh plugin install valkia/dsh-plugin-computer-use
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install valkia/dsh-plugin-computer-use in your terminal to install the plugin inside DeepSeek Harness; the source is hosted at https://github.com/valkia/dsh-plugin-computer-use.
About this plugin
Large language models reason about text but cannot see the windows and buttons on your desktop. dsh-plugin-computer-use closes that gap for DeepSeek Harness: once mounted, the model reads screenshots and accessibility trees through the Open Computer Use MCP runtime and performs GUI actions such as clicking, dragging, scrolling, typing, and key presses.
Approval is enforced per tool call by default (per-call), so every desktop action requires explicit user consent; an authored preset can grant an explicit allow for controlled deployments. The plugin targets macOS 14+, Linux with AT-SPI2, and Windows with UI Automation across arm64 and x64; a single mounted instance serves only one active Session to keep element-index and snapshot namespaces isolated.
Built for developers assembling local desktop automation, assistive workflows, or exploratory GUI interactions: it removes the need to manage MCP process lifecycle, OS permissions, or cross-session conflicts on your own, so a single preset row gives the model hands and eyes for multi-step desktop tasks.
Use Cases
- Read screenshots and accessibility trees to drive multi-step desktop actions like clicking, typing, and dragging
- Control native desktop application windows without relying on browser extensions
- Equip assistive or exploratory GUI tasks with hands and eyes inside an automation pipeline
Best For
- Developers building local desktop-automation pipelines
- Teams integrating LLM-driven GUI control into assistive-interaction projects
- Engineers prototyping Computer Use without maintaining their own MCP infrastructure
Related Plugins
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.