nvidia-kimi-vision

TotalClaw 作者 totalclaw

通过 NVIDIA NIM API 使用 NVIDIA Kimi K2.5 视觉模型分析图像。非常适合为 MiniMax M2.5、GLM-5 等非视觉模型或任何没有本机图像支持的模型添加视觉。支持 png、jpg、jpeg、webp。

安装 / 下载方式

TotalClaw CLI推荐

totalclaw install totalclaw:totalclaw~miladnoo-nvidia-kimi-vision

cURL直接下载，无需登录

curl -fsSL https://skills.taituai.com/api/skills/totalclaw%3Atotalclaw~miladnoo-nvidia-kimi-vision/file -o miladnoo-nvidia-kimi-vision.md

## 概述（中文）

## 原文

# NVIDIA Kimi Vision

Fast image analysis using Kimi K2.5 multimodal model from NVIDIA NIM.

## Why This Skill?

- **Fast** - NVIDIA NIM inference
- **Quality** - Kimi K2.5 is a solid vision model
- **Simple** - Just pass an image and prompt
- **Free tier** - Available through NVIDIA build.nvidia.com

## API Setup (IMPORTANT)

When using this skill, if no API key is found, it will automatically guide the user through setup:

### Step 1: Get a Free API Key
1. Go to **https://build.nvidia.com**
2. Sign up / Log in with GitHub or Google
3. Search for "Kimi K2.5"
4. Click on the model and get your free API key

### Step 2: Save the Key
```bash
# Option A: Save to file (recommended)
mkdir -p ~/.config
echo 'your-api-key-here' > ~/.config/nvidia-kimi-api-key

# Option B: Pass directly when running
python3 scripts/analyze_image.py photo.jpg "What's this?" sk-your-key-here
```

### First Time Setup (for agents)
When a user tries to use this skill without an API key, the script will output clear setup instructions. Guide them through:
1. Visiting https://build.nvidia.com
2. Getting their free API key
3. Saving it to ~/.config/nvidia-kimi-api-key

## Usage

```bash
python3 scripts/analyze_image.py <image_path> "<prompt>" [api_key]
```

### Examples

```bash
# What's in this image?
python3 scripts/analyze_image.py "/path/to/image.jpg" "Describe what's in this image"

# Extract text from screenshot
python3 scripts/analyze_image.py "/path/screenshot.png" "Extract all text"

# Analyze a meme
python3 scripts/analyze_image.py "/path/meme.jpg" "Explain this meme"

# With API key inline
python3 scripts/analyze_image.py photo.jpg "What's this?" sk-xxxxx
```

## Image Formats

Supports: png, jpg, jpeg, webp

## Rate Limits

The free tier through NVIDIA NIM has some limits, but they're not clearly documented on the site. Check https://build.nvidia.com for the latest info on your specific key's limits.