A powerful vision extension for AI agents that enables high-speed image analysis and OCR using the NVIDIA Kimi K2.5 model.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install nvidia-kimi-vision
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install nvidia-kimi-vision using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is NVIDIA Kimi Vision?
NVIDIA Kimi Vision leverages the NVIDIA NIM API to provide high-performance multimodal capabilities to models that lack native vision support, such as MiniMax M2.5 or GLM-5. By integrating this skill, developers can quickly process visual data with minimal latency, making it an essential addition to any Openclaw Skills collection.
This tool bridges the gap between text-only LLMs and complex visual tasks, offering a simple yet robust interface for image interpretation. Whether you are building an automated researcher or a productivity agent, this skill allows for seamless visual context gathering through the highly optimized NVIDIA NIM infrastructure.
NVIDIA Kimi Vision Use Cases
- Describing visual content for text-based AI models like GLM-5 or MiniMax.
- Performing high-accuracy Optical Character Recognition (OCR) on screenshots and digital documents.
- Providing technical explanations for complex diagrams or code snippets captured as images.
- Explaining cultural context or humor within memes and social media assets.
- Enabling visual reasoning for autonomous agents within the Openclaw Skills ecosystem.
How NVIDIA Kimi Vision Works
- The user provides a local image path and a text-based prompt to the execution script.
- The script checks for a valid NVIDIA NIM API key in the local ~/.config directory or via command-line arguments.
- The image file (PNG, JPG, or WebP) is read and transmitted to the NVIDIA Kimi K2.5 vision model endpoint.
- The model analyzes the image content based on the user's specific query.
- The script returns the generated textual analysis or extracted data directly to the user or calling agent.
NVIDIA Kimi Vision Setup
To get started, ensure you have Python 3 installed on your system. You will also need a free API key from the NVIDIA NIM platform.
- Visit build.nvidia.com and sign up.
- Locate the Kimi K2.5 model and generate your API key.
- Configure the skill by saving your key locally:
mkdir -p ~/.config
echo 'your-api-key-here' > ~/.config/nvidia-kimi-api-key
You can also execute the script directly by passing the key as an argument:
python3 scripts/analyze_image.py path/to/image.jpg "Describe this image" sk-your-api-key
NVIDIA Kimi Vision Data Schema & Taxonomy
The skill manages configuration and file inputs to produce structured textual output. Below is the metadata and file handling taxonomy:
| Item | Type | Purpose |
|---|---|---|
| API Key | Authentication | Stored in ~/.config/nvidia-kimi-api-key for persistent access. |
| Image Input | File Path | Supports png, jpg, jpeg, and webp formats. |
| Prompt | String | The specific question or command for the vision model. |
| Analysis Output | Text/Markdown | The response returned by the Kimi K2.5 model via NIM. |
NVIDIA Kimi Vision Advanced Features
- Seamlessly adds multimodal capabilities to non-vision models within Openclaw Skills pipelines.
- Automated setup assistance that guides users through API key acquisition if credentials are missing.
- Low-latency inference powered by NVIDIA NIM (NVIDIA Inference Microservices).
- Support for modern image formats like WebP for web-centric automation workflows.
- Flexible credential management supporting both configuration files and direct CLI injection.
SKILL.md
Loading
Related Openclaw Skills
Visual Essay Engine
A structural content production engine that transforms scattered AI images into coherent, narrative-driven visual essays ready for publication.
Distribution Agent Publisher Pack
A specialized agent skill that transforms images into platform-specific social media content with vision-based music hints and automated routing.
Meeting Note Skill
A high-performance analysis tool that converts collaborative discussions into structured decision tracks, atomic insights, and prioritized action items.
Interview Designer
An AI-powered interview preparation tool that transforms resumes into structured, evidence-based hiring strategies using top-tier methodology.
MBTI Personality Test
A comprehensive 70-question psychological assessment tool to determine your four-letter Myers-Briggs Type Indicator.
MoltHands
MoltHands is a task collaboration platform designed for AI agents to publish requirements, execute tasks, and manage incentives through a points-based system.
Featured*
