Scene Chain Workflow - Qaptly Docs

The Scene Chain is a 4-stage AI processing pipeline that transforms natural language descriptions into automated test executions.

Overview

┌─────────────┐    ┌─────────────┐    ┌─────────────┐    ┌─────────────┐
│   Stage 1   │ →  │   Stage 2   │ →  │   Stage 3   │ →  │   Stage 4   │
│   Actions   │    │   Scene     │    │   Final     │    │  Execution  │
│   Review    │    │   Matching  │    │ Validation  │    │             │
└─────────────┘    └─────────────┘    └─────────────┘    └─────────────┘

Interface Layout

┌────────────────────────────────────────────────────────────────────────────┐
│  AI Assistant                                     [Target: MyApp ▼]        │
├────────────────────────────────────────────────────────────────────────────┤
│                                                                            │
│  Progress:  ●─────●─────●─────○                                            │
│            Stage 1  2    3    4                                            │
│                                                                            │
├────────────────────────────────────────────────────────────────────────────┤
│                                                                            │
│  Session History                                      [≡ History Drawer]   │
│                                                                            │
│  ┌──────────────────────────────────────────────────────────────────────┐  │
│  │                                                                      │  │
│  │  Current Stage Content                                               │  │
│  │  (Varies by stage)                                                   │  │
│  │                                                                      │  │
│  └──────────────────────────────────────────────────────────────────────┘  │
│                                                                            │
├────────────────────────────────────────────────────────────────────────────┤
│  ┌──────────────────────────────────────────────────────────┐ [Ctrl+↵]    │
│  │ Type your test scenario...                               │ [Send]      │
│  └──────────────────────────────────────────────────────────┘             │
└────────────────────────────────────────────────────────────────────────────┘

Starting a Session

Step 1: Select Target

Click the Target dropdown
Select the application to test
This provides context for AI matching

Step 2: Enter Prompt

Type your test scenario in natural language
Be descriptive but concise
Include key actions you want to test

Step 3: Submit

Press Ctrl+Enter to submit
Or click the Send button

Example Prompts

✅ Good: "Log in, go to profile settings, change password"
✅ Good: "Add item to cart and proceed to checkout"
✅ Good: "Search for 'laptop' and filter by price"

❌ Too vague: "Test the app"
❌ Too technical: "POST /api/login with credentials"

Stage 1: Actions Review

Purpose: AI decomposes your natural language into discrete test actions.

What Happens

AI analyzes your prompt
Breaks it into individual actions
Presents the action list for review

Interface

┌────────────────────────────────────────────────────┐
│ Stage 1: Actions Review                            │
│                                                    │
│ Your prompt has been analyzed into these actions:  │
│                                                    │
│ 1. ☑ Open the app                                  │
│ 2. ☑ Navigate to login screen                      │
│ 3. ☑ Enter username                                │
│ 4. ☑ Enter password                                │
│ 5. ☑ Tap login button                              │
│ 6. ☑ Verify dashboard loads                        │
│                                                    │
│ [Edit Actions] [Confirm →]                         │
└────────────────────────────────────────────────────┘

User Actions

Review: Check the action list
Edit: Modify, add, or remove actions
Confirm: Accept and proceed to Stage 2

Stage 2: Scene Matching

Purpose: AI matches each action to existing scenes in your library.

What Happens

AI searches your scene library
Finds matching scenes for each action
Presents matches with confidence levels

Interface

┌────────────────────────────────────────────────────────────────────┐
│ Stage 2: Scene Matching                                            │
│                                                                    │
│ Actions matched to scenes:                                         │
│                                                                    │
│ Action: "Navigate to login screen"                                 │
│ ├─ ✅ Match: "SC-001: Open Login Page" (95% confidence)            │
│ └─ 📝 Alternative: "SC-015: Navigate to Auth" (72% confidence)     │
│                                                                    │
│ Action: "Enter username"                                           │
│ ├─ ✅ Match: "SC-002: Input Username" (98% confidence)             │
│                                                                    │
│ Action: "Verify dashboard loads"                                   │
│ ├─ ⚠️ No exact match found                                         │
│ └─ 📝 Suggestion: Create new scene?                                │
│                                                                    │
│ [Change Matches] [Confirm →]                                       │
└────────────────────────────────────────────────────────────────────┘

Match Indicators

Icon	Meaning
✅	High confidence match
📝	Alternative available
⚠️	No match found

User Actions

Review matches: Check each action’s matched scene
Change match: Select alternative scenes
Handle missing: Address actions without matches
Confirm: Accept and proceed to Stage 3

Stage 3: Final Validation

Purpose: AI validates the complete test sequence for consistency.

What Happens

AI reviews the entire chain
Checks for completeness
Validates execution feasibility
Presents final plan

Interface

┌────────────────────────────────────────────────────────────────────┐
│ Stage 3: Final Validation                                          │
│                                                                    │
│ Execution Plan Summary:                                            │
│                                                                    │
│ ┌────────────────────────────────────────────────────────────────┐ │
│ │ 1. SC-001: Open Login Page                                     │ │
│ │    └─ Dependencies: App launched ✓                             │ │
│ │                                                                │ │
│ │ 2. SC-002: Input Username                                      │ │
│ │    └─ Dependencies: On login screen ✓                          │ │
│ │                                                                │ │
│ │ 3. SC-003: Input Password                                      │ │
│ │    └─ Dependencies: Username entered ✓                         │ │
│ │                                                                │ │
│ │ 4. SC-004: Submit Login                                        │ │
│ │    └─ Dependencies: Credentials entered ✓                      │ │
│ │                                                                │ │
│ │ 5. SC-010: Verify Dashboard                                    │ │
│ │    └─ Dependencies: Logged in ✓                                │ │
│ └────────────────────────────────────────────────────────────────┘ │
│                                                                    │
│ Validation: ✅ Plan is valid and ready to execute                  │
│                                                                    │
│ [Modify Plan] [Execute →]                                          │
└────────────────────────────────────────────────────────────────────┘

Validation Checks

Dependency chains are valid
All required preconditions met
No conflicting actions
Execution order is logical

User Actions

Review plan: Examine the execution sequence
Modify: Make adjustments if needed
Execute: Start Stage 4 execution

Stage 4: Execution

Purpose: Execute the validated test sequence automatically.

What Happens

Execution begins on connected device
Each scene runs in sequence
Progress tracked in real-time
Results collected

Interface

┌────────────────────────────────────────────────────────────────────┐
│ Stage 4: Execution                                                 │
│                                                                    │
│ Executing Scene Chain...                                           │
│                                                                    │
│ ████████████░░░░░░░░░░░░ 50%                                       │
│                                                                    │
│ Current: SC-003: Input Password                                    │
│ Status: ▶️ Running                                                  │
│                                                                    │
│ ┌────────────────────────────────────────────────────────────────┐ │
│ │ ✅ SC-001: Open Login Page              (2.1s)                 │ │
│ │ ✅ SC-002: Input Username               (1.5s)                 │ │
│ │ ▶️ SC-003: Input Password               (running)              │ │
│ │ ⏳ SC-004: Submit Login                                        │ │
│ │ ⏳ SC-010: Verify Dashboard                                    │ │
│ └────────────────────────────────────────────────────────────────┘ │
│                                                                    │
│ [Pause] [Stop]                                                     │
└────────────────────────────────────────────────────────────────────┘

Scene Status

Status	Icon	Description
Completed	✅	Successfully executed
Running	▶️	Currently executing
Pending	⏳	Waiting to execute
Failed	❌	Execution failed

Controls

Pause: Temporarily stop execution
Stop: Cancel remaining execution

On Failure

┌────────────────────────────────────────────┐
│ ❌ Scene Failed                             │
│                                            │
│ Scene: SC-003: Input Password              │
│ Error: Element not found                   │
│                                            │
│ [Retry Scene] [Skip] [Abort Chain]         │
└────────────────────────────────────────────┘

Session Management

Session History

Access past sessions:

Click the History Drawer button (≡)
Browse previous sessions
View status and results
Optionally replay sessions

Session States

State	Description
Active	Currently in progress
Completed	Successfully finished
Failed	Terminated with errors
Cancelled	Manually stopped

Pagination

History includes pagination for many sessions:

Use arrows to navigate pages
Most recent sessions first

Tips for Best Results

Writing Effective Prompts

Be specific: “Log in with email” vs “Log in”
Use app terminology: Match language in your app
Order matters: Describe actions in sequence
Include verification: “Verify success message appears”

Handling Missing Matches

Create scenes first: Build your scene library
Use alternatives: Check suggested alternatives
Manual fallback: Some actions may need manual design

Improving Match Quality

Name scenes descriptively: Clear names help AI matching
Add descriptions: Detailed scene descriptions improve matching
Consistent terminology: Use same terms in prompts and scenes

Related:

​Overview

​Interface Layout

​Starting a Session

​Step 1: Select Target

​Step 2: Enter Prompt

​Step 3: Submit

​Example Prompts

​Stage 1: Actions Review

​What Happens

​Interface

​User Actions

​Stage 2: Scene Matching

​What Happens

​Interface

​Match Indicators

​User Actions

​Stage 3: Final Validation

​What Happens

​Interface

​Validation Checks

​User Actions

​Stage 4: Execution

​What Happens

​Interface

​Scene Status

​Controls

​On Failure

​Session Management

​Session History

​Session States

​Pagination

​Tips for Best Results

​Writing Effective Prompts

​Handling Missing Matches

​Improving Match Quality

Overview

Interface Layout

Starting a Session

Step 1: Select Target

Step 2: Enter Prompt

Step 3: Submit

Example Prompts

Stage 1: Actions Review

What Happens

Interface

User Actions

Stage 2: Scene Matching

What Happens

Interface

Match Indicators

User Actions

Stage 3: Final Validation

What Happens

Interface

Validation Checks

User Actions

Stage 4: Execution

What Happens

Interface

Scene Status

Controls

On Failure

Session Management

Session History

Session States

Pagination

Tips for Best Results

Writing Effective Prompts

Handling Missing Matches

Improving Match Quality