Reviews that only report real bugs
A cheap model lists every possible bug. A stronger model then tries to prove each one wrong. Only the bugs that survive reach your pull request.
Open source and runs on your machine with any model. It double-checks every bug before showing you.
It finds the cause, makes the change, writes a test, and runs it.
Real sessions, replayed in Aster's own interface. Click around, it's the real thing.
Runs in the terminal, the desktop app, VS Code, Zed, the browser, and Telegram.
A cheap model lists every possible bug. A stronger model then tries to prove each one wrong. Only the bugs that survive reach your pull request.
Ask how something works, then let Aster make the change. It can write a plan first and wait for your approval.
Simple rules for what it may read, edit, and run. Every command runs in a sandbox, so your secrets stay hidden.
Aster keeps notes about your project and loads them only when they help. Eighteen skills come built in, and tools made for other agents through the open Agent Skills, Agent Plugins, and MCP standards install as they are.
Every major model through one account and one bill, at cost or as a flat monthly bundle.
Start Aster from a GitHub issue, Slack, or a schedule. It works in a safe cloud machine and opens a checked pull request.
One place for your team's skills, rules, and memory, with single sign-on, audit logs, and spending limits.
Sessions follow you from terminal to desktop, editor, and phone, with the full history and what each task cost.
For anyone, on their own machine
$0forever
Available now
For people who want it to just work
$20/mo
Everything in Open source, plus
For teams that review together
$25/person/mo
Everything in Pro, plus
For companies with policy needs
Custom
Everything in Team, plus
Paid plans are not open yet. Model usage and background agents are billed on top of a plan.
Then run aster in any project. It asks how to connect the first time.