cap_llm_inspect_image

Inspect local image files and generate natural-language descriptions.

2.0k|410|Updated Apr 17, 2026

Search Tags:#visual analysis #local filesystem #on-device inference #image inspection #OCR text spotting #prompt guided #IoT device

Ready to run

npx skills add https://github.com/espressif/esp-claw --skill cap-llm-inspect-image

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: cap_llm_inspect_image
Source: https://github.com/espressif/esp-claw/tree/main/components/claw_capabilities/cap_llm_inspect/skills/cap_llm_inspect_image
Command: npx skills add https://github.com/espressif/esp-claw --skill cap-llm-inspect-image

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Users need quick, accurate visual understanding of what is inside an image stored on their device, without manually opening and interpreting the file.

Core Features & Use Cases

Local image inspection: Analyze a single image from an absolute filesystem path already present on the device.
Prompt-guided visual analysis: Provide a task-specific prompt to control what to look for (objects, scenes, or visible text).
Uncertainty-aware descriptions: Report uncertainty when the image is blurry or the target cannot be reliably identified.

Use case example: When a device receives a photo of a document or scene, the user can ask it to identify key objects and extract any visible text.

Quick Start

Ask the device to run image inspection on the known local path and describe the visible objects and any readable text.

Frequently Asked Questions about cap_llm_inspect_image

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I analyze a local image file on an IoT device?▼

To analyze a local image file on an IoT device, provide an absolute filesystem path and a clear prompt to get a natural-language description of visible objects, scenes, or text. This local image inspection works entirely on-device.

Can I extract text from a local photo using prompt-guided visual analysis?▼

Yes, you can extract text from a local photo using prompt-guided visual analysis. By providing a task-specific prompt targeting OCR text spotting, the Skill reads visible text and reports it within a natural-language description.

Does on-device image inspection work with remote URLs?▼

No, on-device image inspection does not work with remote URLs. The Skill explicitly rejects remote URLs and non-image inputs, requiring an absolute local file path already present on the device filesystem to perform visual analysis.

What happens during image inspection if the target object is blurry?▼

If the target object is blurry during image inspection, the Skill produces uncertainty-aware descriptions. It reports uncertainty when the image is blurry or the specified target cannot be reliably identified from the local file.

What is the best way to interpret a screenshot stored on my device?▼

The best way to interpret a screenshot stored on your device is using local image inspection. You provide the absolute file path and a prompt specifying what to inspect, receiving a natural-language interpretation of the screenshot content.

How do I perform visual question answering on a camera frame?▼

To perform visual question answering on a camera frame, run image inspection on the known local path of the captured frame. Provide a clear prompt asking what to inspect, and the Skill delivers a natural-language answer about the visual content.