Evaluate VLM browser agents with Eval Protocol, Kernel browsers, and Fireworks.
-
Updated
Mar 25, 2026 - Python
Evaluate VLM browser agents with Eval Protocol, Kernel browsers, and Fireworks.
Records a workflow once. Replays it with variations you can edit like code.
Autonomous AI desktop agent powered by local Ollama LLM — automates files, Excel, web, shell and GUI tasks on Windows.
🛠️ 沐挚的 Codex Skill 合集:修复 Windows 端 Computer Use 不可用、清理孤儿会话 | MuZhi Codex skills: fix Windows Computer Use, session cleanup & more
Cinnamon/X11 Computer Use for Codex: a standalone Rust CLI/MCP plugin
Local UI-grounding specialist for hybrid AI agents. Qwen3-VL-2B LoRA. Screenshot + text target → strict JSON bbox. Drop-in for Claude Code, Codex, browser-use. Cuts GPT-4V cost & latency.
Computer use for Linux. An MCP server that lets Claude Code, OpenAI Codex or any MCP client drive a real browser on your own machine, with a live view, human takeover and safety guardrails.
Ambient AI companion — local-first, ever-evolving desktop pet that thinks, learns, acts and grows. Free, open-source, MIT.
Pixel-perfect click coordinates for AI agents. Vision models miss by 30–40 px; OCR + template matching land on the exact pixel. Pure-CPU screen automation — no GPU, no cloud API, no token cost.
Cross-platform LLM computer-use agent — OpenAI, Anthropic, Gemini, NVIDIA NIM and more control your PC behind a permission-safe tool layer.
Agent-guided generation and evaluation of editable presentation repair tasks, informed by real model rollouts.
Computer Use Agent developed with Jev
ZetaTron — a self-directed autonomous AI agent harness. Local-first LLM runtime with memory, tooling, vision grounding, peer mesh, and a real personality.
⚡ Code-driven AI browser automation for Chrome. Compiles natural language into native JavaScript workflows with sub-100ms visual actions & live visual feedback.
Give coding agents a real browser on headless Linux.
Two-tier computer use for Linux: code perceives, Jev decides in ~200ms, Claude only when stuck. Moves the real cursor on Hyprland/Wayland.
LLM discovers a legacy-UI workflow once; it becomes a typed, replayable capability that runs without the model.
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.
Ephemeral, sandboxed virtual desktop environment for full computer-use AI agents with sub-500ms atomic state rollback and Anthropic protocol compliance.
To associate your repository with the computer-use topic, visit your repo's landing page and select "manage topics."