AI Testing Agent

Your AI QA engineer that
writes, runs and maintains tests

ReTibe IDE is an autonomous agent that reads your codebase, plans a test strategy, writes the scenarios, runs them on a schedule, catches regressions and reports back, without being asked each time.

Download for macOS Docs

Apple Silicon and Intel builds, both signed. Windows and Linux are not out yet.

Calendar 125.json Test preview

Passed

Scenarios24

Creator Login & Redirect

Creator Onboarding E2E

Member Discover

Member Home Feed

Member Login

Schedules3

daily 03:00 — auth

weekly — full regression

Live preview

https://app.example.com/discover

Execution log

00:08:01web_navigate — 200 OK1.2s

00:08:02web_fill_form — email, pass0.8s

00:08:03web_click — Submit0.3s

00:08:07running step-4…

step-4 4/10 steps

How it works

Autonomous testing in seven steps

From code analysis to bug report, the agent handles the entire QA cycle.

Understands

Scans your codebase: routes, forms, auth, API endpoints, selectors.

Plans

Generates a test strategy with priorities and categories.

Creates

Writes JSON scenarios with real selectors and URLs taken from your code.

Executes

Runs tests on a schedule: daily, weekly, monthly or custom cron.

Detects

Visual regressions, flaky tests and performance drops.

Reports

Dashboard analytics and detailed reports with screenshots.

Evolves

Updates the docs and suggests new coverage when the code changes.

Agent capabilities

What the agent can do

Thirty-eight agent tools, eleven of them wired straight into the ReTibe platform, 98 editor commands and a codebase it has actually read.

Autonomous test creation

Eleven platform tools write scenarios, checklists, parameters and schedules, from a single prompt or a whole-project batch.

Supervised execution

Watch it work in real time: live browser screenshots, step-by-step logs, streamed over WebSocket into the editor.

Full control when you want it

Inline diff review: accept or reject each change with ⌘Y and ⌘N. Nothing is saved behind your back.

Regression detection

Pixel-level visual comparison, flakiness scored 0 to 100 and baseline management that does not need babysitting.

Coverage intelligence

Checklists generated from scenarios, an execution mode with pass and fail tracking, import and export, coverage analytics.

Deep codebase understanding

An L1 to L4 progressive scan builds project docs the agent loads automatically, up to a 200 000 character context budget.

An agent that works while you sleep

It analyses the project, drafts a test plan, generates scenarios, schedules them and reports what it found. No setup ritual each time.

Reads the project and proposes a test plan for the parts nothing covers

Schedules and runs them itself: daily, weekly, monthly or cron

Eleven platform tools: scenarios, checklists, parameters, schedules and docs

Keeps the project documentation current as the code moves

AI Agent

I found 3 untested routes: /api/payments, /api/orders and /settings. Shall I cover them?

Yes, generate all three and schedule them daily

Three scenarios, 24 steps in total. Scheduled for 03:00 daily.

botfake_save_scenario ×3

botfake_create_schedule ×3

Scenario #307 — 3 changes

5+"action": "security"

6 "action": "seo_analysis"

7 "action": "performance_analysis"

8−"action": "web_screenshot"

8+"action": "web_scroll_screenshot"

9 "action": "mobile_device"

10+"device": "iPhone 13 Pro"

Keep ⌘Y · Undo ⌘N

The whole scenario lifecycle, in the editor

Create, edit, run and review without leaving it. A virtual filesystem makes scenarios stored on the server behave like local files.

A botfake:// filesystem with full read and write access

Inline diff review: keep or undo each AI change with ⌘Y and ⌘N

Environment filter across every artefact type

Scenario history with success rates and failure analysis

Your codebase, documented and always in context

Scan any project and get structured documentation: routes, forms, auth flows, selectors, code excerpts. Stored locally, loaded into the agent's context automatically.

Four scan levels, from an L1 overview to an L4 pass over every reachable source file

A rescan overwrites only files still carrying the autogen marker; delete that line and the file is yours

The agent creates and updates docs through its own tool calls

A 200 000 character context budget, filled in priority order

Project docs

Project docsL4

overview

routes-api

auth

business-flows

testing-strategy

# Business Flows

## User Registration

1. Navigate to /register

2. Fill email, username, password

3. Accept the terms checkbox

4. Click "Create Account"

5. Verify the redirect to /dashboard

Foundation

The editor you already know

Built on VS Code, so the interface, extensions, themes and keybindings are the ones you have. Plus 98 ReTibe commands and the agent's own toolset.

Extensions Monaco editor Integrated terminal Git integration Themes Keybindings AI chat built in Security scanning Analytics and reports

Agent vs manual

Replace hours of manual work with minutes

Task	Manual QA	ReTibe agent
Analyse a new project	Two to three days	One L4 scan
Write a scenario	Hours each	Seconds, from the project docs
Run the regression suite	Someone has to remember	Cron, daily or weekly
Spot a visual bug	The human eye, inconsistently	Pixel diff against a baseline
File a bug report	Typed into Jira by hand	Filed automatically with screenshots
Track suite health	A spreadsheet	Dashboard with flakiness scoring

Download

Pick your platform

macOS

Apple Silicon and Intel

Apple Silicon .dmg Intel .dmg

Windows

x64 / ARM64

Not out yet

Linux