The "computer-use" topic on GitHub groups 31 open-source projects in the DeepSeek Harness (DSH) ecosystem, led by agent-vision-toolkit with 1.2k GitHub stars. agent-vision-toolkit — A better vision toolbox and skill for pure-text models to "see": multi-image understanding, image Q&A, frontend UI restoration, GUI automation and more, with optional seamless integration into multiple mainstream agents, recognizing pasted images directly | A vision toolkit and skill designed for text-only llms — image Q&A, long-screenshot OCR, frontend UI restoration, and GUI automation, with optional seamless integration for Codex, Claude Code, Pi, Oh My Pi, and OpenCode. Every project here is indexed by DSH Universe with live GitHub data — stars, activity and install status — so you can compare and install directly.
A better vision toolbox and skill for pure-text models to "see": multi-image understanding, image Q&A, frontend UI restoration, GUI automation and more, with optional seamless integration into multiple mainstream agents, recognizing pasted images directly | A vision toolkit and skill designed for text-only llms — image Q&A, long-screenshot OCR, frontend UI restoration, and GUI automation, with optional seamless integration for Codex, Claude Code, Pi, Oh My Pi, and OpenCode
Desktop automation MCP server — computer use for any AI agent: control screen, windows, mouse/keyboard, and Chrome via Model Context Protocol (stdio). Not a DSH plugin — DSH users install dsh-nuphus-mcp.
One macOS app for Claude Code, Codex, and every agent runtime you use — scheduled runs, global hotkey launcher, per-run git worktrees, one review board.
Computer control plugin for DeepSeek Harness: fresh Accessibility observations, stale-state rejection, scoped permissions, and safe input (currently supports macos)｜Accessibility-first macOS Computer Use bundle for DSH with fresh observations, stale-state rejection, scoped permissions, and safe input.
Text-first browser & background macOS control for DeepSeek Harness (DSH): target the right process and window without taking the user's pointer. Provides text-first computer control for DSH: operate Chromium and macOS in the background without stealing focus or moving the mouse.
Auditable vision and cross-platform Computer Use runtime for DeepSeek Harness — strict evidence, health-checked failover, original pixels, and Token accounting.
Desktop computer use for DeepSeek Harness: macOS app listing, accessibility-tree window capture, screenshots, and synthesized input, cloned from Codex Computer Use
DeepSeek Harness plugin that does QA: an agent explores your app like a real user and leaves evidence, then the explored path is exported as a deterministic replay scenario you run every release. Orchestrates the browser, macOS desktop, iOS and Android drivers — never a false green.
DeepSeek Harness (DSH) vision plugin — gives the agent screen/window vision + computer-use ability (see/ocr/list_windows + mouse and keyboard control), cross-language calls to the bundled Python cvision, available on Windows / macOS.
A plugin used for the real-device evaluation in “On the Capability Boundaries of Turn-Based GUI Agents in Real-Time GUI Environments: A Human Perception–Action Perspective”
Windows desktop control toolkit for LLM agents: UI Automation with Chromium/Electron accessibility wake-up, verified clicks, and a capture-excluded activity indicator.
DeepSeek Harness plugin for macOS desktop automation. Observes the accessibility tree, gates risky actions behind one-time approval, permanently refuses secure fields, and returns bounded visual evidence — through a native helper you build and grant locally.
