OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Browser Agents

Typesafe Computer Use

Typesafe Computer Use is a Python agent that pursues plain-English goals on a Mac by combining screen OCR, accessibility controls, TypeSafe action classification and selective writing-model assistance.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Typesafe Computer Use — OpenViST project profile in Browser Agents.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you want to explore goal-directed desktop automation on macOS and can review a dry run before allowing actions on the real mouse and keyboard.

Who it is for

  • Mac users exploring plain-language desktop task automation.
  • Developers interested in OCR-driven interaction and classifier-guided computer-use agents.
  • People evaluating a Python project that separates action selection, deterministic state handling and free-text generation.

Documented uses

  • Translate a written desktop goal into a sequence of screen-aware actions.
  • Read text and labelled controls from the frontmost window to inform the next action.
  • Enter free text when a task requires it, using a writing model where configured.
  • Inspect a captured screen and payload with the documented inspection command.
Before you choose

Requirements & limits

The repository documents macOS 14 or newer and Python 3.12 or newer for installation, with uv required. Windows 10 and 11 are marked experimental. A TypeSafe API key is required; an Anthropic API key is optional for specified writing-model functions. MIT license.

  • Live mode acts through the real mouse and keyboard; the project recommends beginning with a dry run.
  • Screen Recording and Accessibility permissions are required for useful capture and synthetic clicks.
  • A TypeSafe API key is needed for every decision; writing-model use is optional and limited to documented functions.
  • The README describes the project as beta and under heavy development.
  • The project’s design relies on deterministic code to supply facts such as dates and focus state, rather than asking the classifier to infer all context from pixels.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up Typesafe Computer Use on macOS

Prepare the repository and credentials for a first dry-run command.

Before you start

  • macOS 14 or newer
  • Python 3.12 or newer
  • uv
  • A TypeSafe API key
  • Screen Recording and Accessibility permissions for the terminal

Steps

  1. Clone the repository and change into its directory.

  2. Run the command below to install the project environment.

    Documented command
    uv sync

  3. Copy .env.example to .env at the repository root and provide TYPESAFE_API_KEY.

  4. Grant the terminal Screen Recording and Accessibility permissions in System Settings > Privacy & Security.

  5. Run the command below to make a dry run that prints what it would do.

    Documented command
    uv run clicker "open the Playground"

Expected outcome

The project is set up and the documented first-run dry run prints a proposed action without driving the computer.

Guide limitations

  • The dry run is limited to one step.
  • The project is beta and warns that settings and behavior can change between 0.x releases.
  • Live interaction requires --act and controls the real mouse and keyboard; the README recommends starting with a dry run.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
11 October 2026
Original website title
awlevin/typesafe-computer-use · Official homepage title · retained in its source language
How we review sourcesSuggest a correction