Document Tools
Note
These tools are Careti-exclusive features.
Careti provides powerful tools that allow AI to directly analyze images and documents. Thanks to these tools, AI can read and analyze files just by their path, without users having to manually attach them.
Image Analysis Tool (analyze_image)
The analyze_image tool analyzes image files to describe their content or answer specific questions.
Use Cases
- UI/UX Review: Check for layout issues, alignment errors, visual bugs in screenshots
- Text Extraction: Extract text from screenshots, error dialogs, logs
- Code Review: Analyze code screenshots and identify patterns
- Design Comparison: Compare UI implementations with design mockups
- Error Analysis: Interpret error dialogs and stack traces
Supported Formats
- PNG, JPG, JPEG, WebP, GIF, AVIF, BMP, TIFF
Usage
User: Extract the error message from this screenshot (./screenshots/error.png)
AI: [Uses analyze_image tool] → Analyzes image → Provides extracted error message
Requirements
- Careti Account Login Required: Image analysis is processed through the Careti API
- Automatic Image Optimization: Large images are automatically optimized before sending
Settings
In Settings → Features, you can select the Image Analysis Model:
- Gemini 3.0 Flash Preview (default)
- Gemini 2.5 Flash
Document Reading Tool (read_document)
The read_document tool reads various document formats and extracts text.
Comparison with Cline
| Feature | Cline | Careti |
|---|---|---|
| Read PDF | User attachment only | AI reads directly by path |
| Read DOCX/XLSX | User attachment only | AI reads directly by path |
| HWPX (Korean Hangul) | ❌ Not supported | ✅ Supported |
| HWP 5.0 (Legacy Hangul) | ❌ Not supported | ✅ Supported |
| PPTX | ❌ Not supported | ✅ Supported |
Supported Formats
| Format | Extension | Description |
|---|---|---|
.pdf | Adobe PDF documents | |
| Word | .docx | Microsoft Word 2007+ |
| Excel | .xlsx | Microsoft Excel 2007+ |
| PowerPoint | .pptx | Microsoft PowerPoint 2007+ |
| Hangul (New) | .hwpx | Hancom Office Hangul 2014+ |
| Hangul (Legacy) | .hwp | Hancom Office Hangul 5.0 |
| Jupyter | .ipynb | Jupyter Notebook |
Unsupported Formats
The following legacy formats are not supported due to technical limitations:
| Format | Extension | Alternative |
|---|---|---|
| PowerPoint 97-2003 | .ppt | Convert to .pptx |
| Word 97-2003 | .doc | Convert to .docx |
| Excel 97-2003 | .xls | Convert to .xlsx |
Tip
Convert legacy formats using LibreOffice, Google Docs, or Microsoft Office.
Usage
User: Analyze docs/spec.pdf in this project
AI: [Uses read_document tool] → Extracts PDF content → Provides analysis
Security
- Path Validation: Path normalization to prevent Path Traversal attacks
- File Size Limit: Maximum 50MB
- Format Validation: Only supported extensions allowed
Limitations
- No Images/Charts: Images and charts within documents are not extracted
- Layout Loss: Tables, multi-column layouts are converted to simple text
Usage Tips
Image Analysis Usage
# Check for UI bugs
"Check if there are layout issues in the ./screenshots/ui-bug.png screenshot"
# Analyze error messages
"Analyze the cause of the error from this screenshot (./error.png)"
# Design review
"Compare if the ./mockup.png design matches the current implementation"
Document Analysis Usage
# Analyze specifications
"Read docs/api-spec.pdf and list the required API endpoints"
# Data analysis
"Summarize the contents of data/report.xlsx"
# Korean document processing
"Extract the key clauses from contracts/contract.hwp"