Installs into .claude/skills of the current project.
Are you the author of Screenshot Analyze?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/davidroliverba-screenshot-analyze)
---
context: fork
---
# /screenshot-analyze
Perform comprehensive analysis of screenshots using Sonnet sub-agents for detailed visual inspection.
## Usage
```
/screenshot-analyze <image-path>
/screenshot-analyze +Attachments/error-screenshot.png
/screenshot-analyze +Attachments/ui-mockup.png --context "login flow"
```
## Instructions
This skill uses **Sonnet model sub-agents** for thorough image analysis with vision capabilities.
### Phase 1: Image Loading
1. Verify the image exists at the specified path
2. Check file type is supported (PNG, JPG, JPEG, GIF, WebP)
3. Note any context provided by user
### Phase 2: Comprehensive Analysis (Sonnet Sub-Agents)
Launch these sub-agents using `model: "sonnet"`:
**Agent 1: Content Extraction** (Sonnet)
```
Task: Extract all visible content from the image
- Read the image file using the Read tool
- Extract all visible text (OCR)
- Identify UI elements, buttons, labels
- Note any error messages or notifications
- Capture data values, IDs, or codes visible
Return: Complete text extraction with location context
```
**Agent 2: Visual Structure Analysis** (Sonnet)
```
Task: Analyse the visual structure and layout
- Read the image file
- Identify the type of content (UI, diagram, document, error, etc.)
- Map the visual hierarchy and layout
- Note colour schemes, branding elements
- Identify interactive elements
- Assess visual quality and clarity
Return: Structural analysis with element inventory
```
**Agent 3: Context & Purpose Analysis** (Sonnet)
```
Task: Understand the purpose and context
- Read the image file
- Determine what application/system is shown
- Identify the workflow or process stage
- Note any problems or issues visible
- Assess what action might be needed
- Connect to known YourOrg systems if recognisable
Return: Contextual interpretation with system identification
```
### Phase 3: Compile Report
```markdown
# Screenshot Analysis
**Image**: {{filename}}
**Analysed**: {{DATE}}
**Type**: {{UI/Error/Diagram/Document/Other}}
## Summary
{{2-3 sentence overview of what the image shows}}
## Extracted Content
### Text Found
{{all readable text, organised by location}}
### UI Elements
| Element | Type | Label/Text | State |
|---------|------|------------|-------|
{{element inventory}}
### Data Values
{{any IDs, codes, numbers, or data visible}}
## Visual Analysis
### Layout Structure
{{description of visual hierarchy}}
### Application/System
- **Identified as**: {{system name if known}}
- **Screen/Page**: {{specific view or page}}
- **Context**: {{workflow stage}}
## Observations
### Key Findings
{{important observations}}
### Issues Detected
{{any errors, warnings, or problems visible}}
### Recommendations
{{suggested actions based on analysis}}
## Related Notes
{{suggest links to related vault notes if system is recognised}}
```
### Notes
- Sonnet provides superior visual analysis capabilities
- Works with screenshots, UI mockups, error messages, dashboards
- Can recognise YourOrg systems like SAP, MROPlatform, Confluence, etc.