Abkus Studio
Login

Autonomous AI agent IDE

Abkus Studio

Autonomous AI agent studio for developers — pair your daemon, edit code, and ship with Hub-routed models.

Subscribe via Abkus Hub View sample QA report
View pricing Sign in
Abkus Studio — agent run

Explorer

src/agent/run.ts

src/tools/qa.ts

src/daemon/pair.ts

src/routes/+page.svelte

package.json

const agent = createStudioAgent();

agent.pairDaemon();

agent.runQa(routes: ['/dashboard','/projects']);

$ abkus-code pair --token ********

daemon connected — macbook-pro.local

→ Running intelligent QA on 6 routes…

4 findings recorded · report finalized

Video

See it in action

A quick walkthrough of the Abkus Studio landpage — capabilities, the intelligent QA sample, pricing and FAQ.

Capabilities

Everything you need to ship with an AI agent

One workspace that pairs your machine, routes models through the Hub, and closes the QA loop for you.

Daemon pairing

Connect abkus-code on your machine to read and edit real files over the network.

Intelligent QA

The agent drives a real browser like a human tester and records findings per route.

Autopilot cloud builds

Spin up cloud projects and let agents build in an Abkus sandbox — no local daemon needed.

Hub-routed models

Use any LLM provider configured on the Abkus Hub. No keys in your code or env.

Runs memory

Every loop is journaled with turns, tools and cost, so agents can plan fixes and retest.

Dual runtime

Run in the browser or on a local daemon — switch runtime per project.

How it works

From idea to tested code in minutes

Three steps, no config files, no API keys in your app.

01

Pair your computer

Install abkus-code and pair the daemon so the agent can read and edit your files.

02

Describe the app

Tell the agent what to build or fix. It writes a plan and implements it.

03

Run intelligent QA

The agent browses like a human, records findings, and loops fix → retest.

Build with agents

Describe the app you want

Tell Studio what to build. We create a cloud project, write a plan, and agents can implement it in an Abkus sandbox — no PC daemon required for cloud projects.

Continue in Studio

Sample

Intelligent QA — see the differential

Abkus Studio drives a browser like a human tester: navigate, screenshot, judge UI, record findings, and close the loop with fix → retest. Below is a sanitized sample report — no live credentials, no secrets.

  • Explores public pages and mobile viewports (390×844)
  • Records bugs and improvements with severity and route
  • Feeds Runs memory so agents plan fixes and retest

Sample intelligent QA report

Sample · sanitized

Sample intelligent QA report (sanitized). Public pages explored; one improvement suggested for mobile CTA touch target; console clean on privacy.

Findings

  • sample-defect-1 Severity: Low Kind: improvement

    Tap Subscribe on mobile

    Route: /landpage#pricing

    Expected: Touch target ≥ 44×44px. Observed: CTA height ~36px on 390px viewport.

  • sample-defect-2 Severity: Medium Kind: bug

    Open free tool card then Back

    Route: /tools

    Expected: Focus returns to the card that was opened. Observed: Focus jumps to document body after Back.

Subscribe to run QA on your apps Help & FAQ Download manual (PDF)

This page never asks for passwords or API keys.

Pricing

Simple Hub billing

One Abkus Hub account. Subscribe to Studio Pro, pair your daemon, and run intelligent QA with Hub-routed models. Checkout and invoices live on the Hub.

Studio Pro

$49 / month

  • Browser IDE + local abkus-code daemon
  • Intelligent QA loops with Runs memory
  • Hub SSO, Hub AI routing, 14-day refund guarantee
Coming soon

14-day unconditional refund on Hub. Cancel anytime.

Prices are charged in USD. BRL amounts are approximate (≈) and may vary with the daily exchange rate.

FAQ

Common questions

Do I need a local daemon?

No. Cloud projects run in an Abkus sandbox with no daemon. Pair a daemon only when you want the agent to touch files on your own machine.

Which models can I use?

Any LLM provider configured on your Abkus Hub account. Studio routes requests through the Hub, so no model keys ever touch your app.

Local vs cloud runtime?

Use the browser runtime for cloud projects and demos. Use the local daemon when you need full filesystem access on your machine.

What if I want to cancel?

Every subscription has an unconditional 14-day refund guarantee on the Hub. Cancel anytime.

Start shipping with an AI agent

Pair your daemon, describe the app, and let intelligent QA close the loop.

Coming soon