More screenshots
Works with the AI Agents you already use
omma — give your agent sight
omma — give your agent sight Turn any screenshot into structured text: OCR with positions, colour palette, layout regions. Redact names and faces before anything leaves your machine
Free
omma — give your agent sight
Example session with this skill installed
Analyze the screenshot at ./logs/error_report.png. I need the text from the red error box at the top and the overall color theme to check if it is in dark mode.
- Read your context and instructions
- Compiled the omma
- Generated the image
I have analyzed the image at ./logs/error_report.png. The OCR process identified the error message, and the spatial color grid confirms the UI state.
Image Analysis Results
| Feature | Value | Details |
|---|---|---|
| Primary Text | Unhandled Promise Rejection: Timeout at /api/auth | Found in Top Region (0, 0, 800, 120) |
| OCR Confidence | 0.94 | High confidence on text extraction |
| UI Theme | Dark Mode | Background grid shows 92% dark gray/black (Hex: #121212) |
| Color Highlights | Red (#FF0000) | Detected in the top-left quadrant |
| Layout | 3 Regions | Header, Main Content, Footer sidebar detected |
OCR Details (Block Level)
{
"ocr": {
"text": "Critical Error: 500\nUnhandled Promise Rejection: Timeout at /api/auth\nCheck logs for details.",
"blocks": [
{"box": [10, 10, 300, 50], "text": "Critical Error: 500", "confidence": 0.98},
{"box": [10, 60, 500, 110], "text": "Unhandled Promise Rejection...", "confidence": 0.94}
]
},
"colors": {
"dominant": ["#121212", "#FF0000", "#FFFFFF"],
"is_dark": true
}
}
Next steps
- Review the
/api/authendpoint for recent latency spikes or timeout configuration changes. - Verify if the red banner styling matches the design spec for high-priority errors.
- Compare the detected OCR text against the known
error_codes.jsonfor mapping.
error_report.png
PNG · 1536×1024
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
Extracting data from images often forces a choice between slow vision models or inaccurate text scrapers. This skill bridges that gap by providing a structured bridge to the contents of screenshots, charts, and diagrams. It delivers precise OCR, spatial layout analysis, and color grids without the overhead of full vision processing unless specifically requested.
What it does
- Structured OCR extracts text with associated bounding boxes, confidence scores, and segmentation data.
- Layout mapping identifies regions in reading order to preserve the original document or UI structure.
- Color analysis generates a palette and a 64x36 spatial color grid to detect UI themes or banner positions.
- Selective redaction allows masking specific X,Y coordinates before processing to protect sensitive data.
- Vision integration provides descriptive analysis when a model is configured, responding to specific prompts.
How it works
- Initialize analysis by providing a file path or base64 data to the
analyze_imagetool. - Configure switches to enable or disable OCR, color grids, or layout detection based on the task.
- Define regions to crop the analysis to a specific quadrant or redact private information.
- Parse structured output using the returned blocks, lines, and confidence intervals to inform responses.
Frameworks & tools
This tool relies on Tesseract for optical character recognition and local image processing for spatial and color analysis. It supports multiple languages and segmentation modes.
Why this beats prompting it yourself
Generic vision prompts often hallucinate coordinates or miss small text in dense UIs. This skill provides raw, structured data including confidence scores and precise pixel boxes that manual prompting cannot consistently replicate.
Use cases
- Debugging system errors from user-provided screenshots by reading logs and stack traces.
- Extracting specific numerical data from complex charts or tabular image formats.
- Determining UI states, such as dark mode or notification presence, via color grid analysis.
- Converting scanned documentation into structured text while maintaining reading order.
Known limitations
OCR functionality requires Tesseract to be installed on the host machine. Vision descriptions are only available if a compatible model is configured.
Requires Tesseract for OCR functionality. Vision descriptions require an additional configured model. OCR results are treated as untrusted data.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 3 days ago
- Free with an Agensi account
Requires
Permissions
File Scopes
Terminal is used to start the local omma MCP server. Read Files is how it opens the image you point it at. Nothing is uploaded: OCR, colours and layout are computed on your machine.