computeruseaisdk
build ai agents fast! hello world MCP server template that controls computer with basic SDK primitives
mcp server
ai function calling for all sdk computer control primitives
launch apps
programmatically launch any application on your system through simple api calls
read content
access and extract text and ui elements from any application window
click interaction
perform precise mouse clicks on any ui element or screen coordinate
text input
enter text into any field or application with full unicode support
key commands
execute keyboard shortcuts and key combinations to control applications
open source alternative
you've heard of openai's operator, you've heard of claude's computer use. now there's an open source alternative:
- fully open source and customizable
- runs locally on your machine
- build your agents in seconds, not hours
- use any llm of your choice
simple to get started
build your own agents starting with our hello world template:
- install our mcp server with a single command
- use our client sdk in typescript, or call via http
- start building agents with hello world template
what will you build?
linkedin assistant
automate browsing linkedin, sourcing leads, and exporting them directly to your google sheets
voice control agent
build agents that respond to voice commands to control applications and perform tasks hands-free
remote ide access
use development tools like cursor ide from your mobile device by creating a remote control interface
google sheets agent
automatically export and organize data from any application directly into structured spreadsheets
workflow sequences
create complex multi-step automation sequences like finding contacts and sending personalized emails
testing agent
develop automated ui testing bots that validate application behavior across different scenarios
reminder assistant
build agents that manage apple reminders, create tasks, and organize your schedule automatically
it helpdesk bot
create support agents that diagnose and solve common it problems without human intervention
frequently asked questions
what is the model context protocol (mcp)?
think of it like a usb-c port for ai applications — a standardized way to connect ai models to different data sources and tools.
what is ai sdk?
pre-made components that makes any llm handle all sdk/api calls
do i need to be a programmer to use this?
we provide a no-code agent builder
which platforms are supported?
currently macos only, with windows and linux support planned for future releases.
built in customizeable chain of thought loop
our sdk includes a proven cyclical reasoning framework that helps your agents continuously adapt:
establish overall task
define clear goals before taking any action
understand context
analyze current location and application state
build a high-level plan
create a sequence of steps before execution
compare available actions
evaluate possible interactions against the plan
maintain state
track progress and adapt to changing conditions
agents continuously cycle through these steps, refining their approach with each iteration
log output provides visibility into each reasoning phase
practical examples
get text from whatsapp
extract conversation text and messages from whatsapp for analysis or archiving
interact with messages app
find interactable elements, type messages, and send them automatically
open arc and scroll pages
launch browser, navigate to sites, and perform scrolling interactions
pricing plans
sdk + mcp server/client
$0
- source code
- chained tools to automate computer and browser
- community support on discord
- build it yourself
no-code agent builder
$199
- prompt llms to create new tools
- save tools and run them on schedule
- 30min 1x1 onboarding call with founder