SimulateInput

English | 中文

SimulateInput is a cross-platform desktop and browser automation platform for testing your own websites, desktop applications, installers, and system-level UI flows.

It combines direct input execution, multiple locator strategies, CLI and MCP interfaces, and YAML-driven reusable test cases so the same automation core can be used by engineers, CI pipelines, and AI agents.

• Cross-platform driver architecture for Windows, macOS, Linux X11, and Linux Wayland compatibility
• Multiple locator strategies:
• structured accessibility lookup
• visible text lookup
• OCR-based text lookup
• image template matching
• coordinate fallback
• Real input actions:
• click
• drag
• type text
• press key
• hotkey
• clear text
• screenshot
• MCP server for AI tool calling
• YAML case runner for repeatable automation flows
• Skill definitions and references for AI-assisted execution

• Windows: primary implementation, real execution and smoke tested
• macOS: MVP driver implemented, requires Accessibility / Automation / Screen Recording permissions
• Linux X11: MVP driver implemented, depends on wmctrl, xdotool, screenshot helpers, and optional AT-SPI tooling
• Linux Wayland: compatibility layer, helper-tool dependent and not yet full parity

• src/simulateinput/
• core engine
• platform drivers
• locators
• CLI
• MCP server
• case runner
• docs/automation-platform-design.md
• architecture and implementation plan
• docs/cross-platform-installation.md
• platform setup, dependencies, and permissions
• skills/simulateinput/
• skill definition and CLI / MCP references
• tests/
• unit tests and smoke case YAML files

$env:PYTHONPATH='src'
python -m simulateinput.cli.main doctor
python -m simulateinput.cli.main session start
python -m simulateinput.cli.main mcp tools