Report Abuse

Basic Information

Terminator is an open-source, accessibility-first desktop automation SDK and agent platform from Mediar.ai designed to record, pre-train and run deterministic workflows across desktop applications and browsers. The project provides language bindings and tools so developers can programmatically control apps, locate UI elements, simulate clicks and typing, manage windows and capture screens. It is cross-platform with the most complete feature set on Windows and partial support on macOS and Linux. Terminator emphasizes recording human workflows into stable, reusable automation code, calling AI only for recovery, and ships with components like a workflow recorder, MCP agent, and bindings for Python, TypeScript/Node.js and Rust. The project is MIT‚Äëlicensed and aimed at developers and teams who want high‚Äëaccuracy, low‚Äëlock‚Äëin automation for both experimentation and production use.

Links

Categorization

App Details

Features
Core automation primitives for locating UI elements by accessibility attributes and performing actions such as click, type and keypress. Cross-platform language bindings including terminator.py (Python), terminator.js (TypeScript/Node.js) and terminator-rs (Rust). Workflow recording and YAML workflow generation for deterministic automations and faster replay. MCP agent integration to orchestrate AI-guided recovery and higher-level decision making. Tools and guidance for inspecting accessibility trees on Windows, macOS and Linux to build reliable selectors. Platform support matrix that marks Windows as most feature-complete and lists partial support for macOS and Linux. Installation and usage examples for programmatic control and editor/install badges for VS Code, VS Code Insiders and Cursor.
Use Cases
For developers Terminator accelerates building reliable desktop automations by turning recorded human interactions into deterministic code and providing APIs to control applications programmatically. For teams it enables rapid process mapping, deployment of AI-assisted execution at scale and reduced manual work without complex legacy RPA tooling. The recorder plus MCP agent model reduces reliance on constant AI calls by pre-training workflows and invoking models only for recovery, which the README claims improves speed and success rates. Cross-language SDKs let engineering and automation teams integrate automations into scripts, CI or desktop tooling. The MIT license and open-source model facilitate experimentation, forking and on‚Äëpremises use with no vendor lock-in.

Please fill the required fields*