🧪 FusionTest

AI-powered UI testing for mobile and desktop — built by FusionLeap.io

FusionTest uses large language models to execute UI tests the way a human would — reading screens, making decisions, and adapting to UI changes automatically. No brittle scripts. No maintenance overhead. Just natural language test goals that run anywhere.

✨ Features

Feature	Mobile (iOS/Android)	Desktop (Win/Mac/Linux)
Natural language test goals	✅	✅
Zero-maintenance execution	✅	✅
Multi-locale / 50+ languages	✅	✅
Hallucination guardrails	✅	✅
Loop detection & backtrack	✅	✅
CI/CD integration	✅	✅
Vision model fallback	—	✅
Cross-app flow testing	—	✅
Electron / Playwright support	—	✅

🏗 Architecture

fusiontest/
├── core/           # LLM action model, embeddings, screen parser
├── mobile/         # iOS (XCTest) + Android (UIAutomator2) adapters
├── desktop/        # Windows (WinAppDriver), macOS (Accessibility API), Playwright
├── guardrails/     # Hallucination handler, loop detector, backtrack engine
└── reporting/      # Dashboard data, Slack/email alerts, CI artifacts

🚀 Quickstart

Prerequisites

Python 3.10+
Node.js 18+ (for Playwright/Electron targets)
Appium 2.x (for mobile)

Install

git clone https://github.com/mohigo/fusion-test.git
cd fusion-test
pip install -e ".[dev]"

Set your API key

export FUSIONTEST_API_KEY=your_key_here
# or copy .env.example → .env and fill in values

Run your first test

fusiontest run --goal "Log in, navigate to settings, update profile name" \
               --platform android \
               --app path/to/app.apk \
               --locale en-US

Full guide: docs/getting-started.md — covers all platforms (web, Android, iOS, Electron, Windows, macOS), Python API usage, configuration, dashboard, CI setup, and tips for writing good goals.

📝 Writing Test Goals

FusionTest accepts natural language goals. No code required.

# tests/goals/onboarding.yaml
name: User onboarding flow
platform: android
locale: en-US
goals:
  - "Open the app and tap Get Started"
  - "Enter email address and password on the signup screen"
  - "Verify the home dashboard loads with a welcome message"

Run it:

fusiontest run --file tests/goals/onboarding.yaml

🔌 CI/CD Integration

GitHub Actions

- name: Run FusionTest
  uses: fusionleap/fusiontest-action@v1
  with:
    goal-file: tests/goals/smoke.yaml
    platform: android
    api-key: ${{ secrets.FUSIONTEST_API_KEY }}

Bitrise / Fastlane

See docs/ci-integrations.md for step-by-step guides.

📊 Reporting

Every test run produces:

Pass/fail summary with stability score
Screen-by-screen action log
Video recording of the test session
Bug report with screenshot + repro steps (on failure)

Results are posted to your FusionTest dashboard and optionally to Slack.

🛠 Configuration

# fusiontest.config.yaml
model:
  backbone: mpnet-base          # or: gpt-4o, claude-sonnet (for MVP/fallback)
  precision_threshold: 0.97
  max_retries: 3

guardrails:
  loop_detection: true
  loop_window: 5                # check last N screen states
  backtrack_on_invalid: true
  max_backtracks: 2

devices:
  android:
    farm: browserstack           # or: aws-device-farm, local
  desktop:
    vision_fallback: true        # enable for non-accessible UIs
    playwright_for_electron: true

reporting:
  slack_webhook: ${SLACK_WEBHOOK}
  save_recordings: true
  dashboard_url: https://app.fusionleap.io

🗺 Roadmap

Core LLM action model (MPNet / GPT-4o / Claude backends)
Android adapter (UIAutomator2 + Appium)
iOS adapter (XCUITest + Appium)
Desktop adapter — Playwright / Electron (with vision fallback)
Desktop adapter — Windows (WinAppDriver / UIA)
Desktop adapter — macOS (NSAccessibility + PyObjC)
Guardrails engine (invalid actions, loops, backtrack)
Goal completion verifier (LLM + heuristic fallback)
Video recording — Playwright WebM→MP4, Appium, Windows screenshot stitch
Reporting — JSON, Slack webhooks, GitHub Actions annotations
CLI — fusiontest run / init / demo / report
MPNet fine-tuning pipeline (fusiontest/training/)
FusionLeap cloud dashboard (React + Recharts + SaaS pages)
Dashboard backend API (FastAPI + PostgreSQL/Supabase)
SaaS backend — auth, org management, API keys, webhooks, billing
Marketing website (Next.js, website/)
Integration tests — runner, guardrails, goal verifier
CI/CD integration guides (docs/ci-integrations.md)
Autonomous Test Discovery — auto-crawl website → generate goals → execute
RAG test library (few-shot goal → flow)
Multi-locale parallel execution
Cross-app flow testing (desktop)

🤝 Contributing

See CONTRIBUTING.md. We welcome issues, PRs, and test goal examples from the community.

📄 License

MIT © FusionLeap.io

Metadata

Release files for fusiontest 0.2.3

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for fusiontest 0.2.3
File	Size	Uploaded
fusiontest-0.2.3.tar.gz	85.9 kB	Sep 27, 2026	Details

Built distribution (wheel)

Table of built distributions (wheels) for fusiontest 0.2.3
File	Interpreter	ABI	Platform	Reset
fusiontest-0.2.3-py3-none-any.whl 96.9 kB Sep 27, 2026	Python 3	none	any	Details

Total release size: 182.8 kB

Release files / fusiontest-0.2.3.tar.gz

Download URL	fusiontest-0.2.3.tar.gz
Size	85.9 kB
Tags	Source
SHA-256 checksum How to use checksums	`acb3d7d138fa9666d70a1cd9f9a179ac2969a6356f0b5640d32c1ea4e8dd3fd9`
BLAKE2b-256 checksum How to use checksums	`7ed3b564d775591471e34d84d40d81a2a1a2e9468a425d3d36c770f7178fa936`
Upload date	Sep 27, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.13.14`

Release files / fusiontest-0.2.3-py3-none-any.whl

Download URL	fusiontest-0.2.3-py3-none-any.whl
Size	96.9 kB
Tags	Python 3
SHA-256 checksum How to use checksums	`ed5add41313e0cfb328b51db5d344daa1e406ade41978201fbfa57bc308b91f8`
BLAKE2b-256 checksum How to use checksums	`695394cefa642debc4aee97aeeb56a401624db73b1fb7ff1073da9c30085f9df`
Upload date	Sep 27, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.13.14`