smart-image-finder（智能图像查找器）

Smart Image Finder

A complete solution for AI agents to find high-quality images for articles, dashboards, and content creation.

No browser required — all methods work via command line in the background.

Core Methods

Method 1: News Website High-Res Extraction (No Browser)

Workflow: curl + grep (Recommended)

Extract image URLs directly from HTML without opening a browser:

# Step 1: Fetch page HTML and extract image URLs
curl -sL "https://www.reuters.com/world/article-slug/" | \
  grep -oE 'https://www\.reuters\.com/resizer/v2/[^"]+\.jpg[^"]*' | head -3

# Step 2: Download with HD parameters
curl -sL -o photo.jpg "https://www.reuters.com/resizer/v2/xxx.jpg?width=3000&quality=100"

# Step 3: Verify it's actually an image
file photo.jpg  # Should show: JPEG image data

Source-Specific Extraction Patterns

# Reuters
curl -sL "$URL" | grep -oE 'https://www\.reuters\.com/resizer/v2/[^"]+\.jpg[^"]*' | head -3

# TechCrunch
curl -sL "$URL" | grep -oE 'https://techcrunch\.com/wp-content/uploads/[^"]+\.(jpg|png)' | head -3

# Bloomberg
curl -sL "$URL" | grep -oE 'https://assets\.bwbx\.io/images/[^"]+\.(jpg|webp)' | head -3

# BBC
curl -sL "$URL" | grep -oE 'https://ichef\.bbci\.co\.uk/news/[0-9]+/[^"]+\.jpg' | head -3

# France24
curl -sL "$URL" | grep -oE 'https://s\.france24\.com/media/display/[^"]+\.jpg' | head -3

# Spaceflight Now (WordPress)
curl -sL "$URL" | grep -oE 'https://spaceflightnow\.com/wp-content/uploads/[^"]+\.jpg' | head -3

# ThePaper (澎湃)
curl -sL "$URL" | grep -oE 'https://imagepphcloud\.thepaper\.cn/pph/image/[^"]+\.jpg' | head -3

Verified Sources (Direct Download Works)

Tech/AI Category

Politics/International Category

Space/Science Category

Chinese Sources

Sources Requiring Manual Save

HD Parameter Quick Reference

# Reuters
?width=3000&quality=100

# TechCrunch
?w=2048

# BBC (replace number in URL)
/1024/  # Max available

# France24
w:1920

# NBC News (URL parameter)
t_fit-1500w

Method 2: Brave Image Search (No Browser)

Best for: Quick general-purpose images, automated workflows.

Basic Usage

# Set API Key (first time)
export BRAVE_API_KEY="your_api_key"

# Search images
curl -s "https://api.search.brave.com/res/v1/images/search?q=keyword&count=10" \
  -H "X-Subscription-Token: $BRAVE_API_KEY" | jq '.results[] | {title, url: .properties.url, width: .properties.width}'

# Get first image URL directly
curl -s "https://api.search.brave.com/res/v1/images/search?q=SpaceX%20Starship&count=1" \
  -H "X-Subscription-Token: $BRAVE_API_KEY" | jq -r '.results[0].properties.url'

Rate Limiting ⚠️

Brave API has rate limits. When making multiple searches:

# Add delay between requests (1-2 seconds recommended)
for query in "query1" "query2" "query3"; do
  curl -s "https://api.search.brave.com/res/v1/images/search?q=$query&count=1" \
    -H "X-Subscription-Token: $BRAVE_API_KEY" | jq -r '.results[0].properties.url'
  sleep 2  # Wait 2 seconds between requests
done

Best practices:

• Free tier: ~1 request/second max
• Add sleep 1 or sleep 2 between batch requests
• If you get 429 (rate limited), wait 60 seconds before retrying

Search Tips

1. Add year: SpaceX Starship 2025 is more precise than SpaceX Starship
2. Add event terms: Keir Starmer Xi Jinping Beijing handshake beats UK China meeting
3. Site restriction: site:reuters.com Trump tariff limits to specific source

One-liner: Search and Download

# Search, get first result URL, download
IMG_URL=$(curl -s "https://api.search.brave.com/res/v1/images/search?q=SpaceX%20launch%202025&count=1" \
  -H "X-Subscription-Token: $BRAVE_API_KEY" | jq -r '.results[0].properties.url')
curl -sL -o spacex.jpg "$IMG_URL"
file spacex.jpg

Verify Image Accessibility

# Check if URL returns 200
curl -sI "$IMAGE_URL" | head -1
# Should return: HTTP/2 200

# Check content type
curl -sI "$IMAGE_URL" | grep -i content-type
# Should return: content-type: image/jpeg or image/png

Method 3: AI Image Generation (No Browser)

Best for: Concept images, creative illustrations, when real photos unavailable.

Free Option: Pollinations

# Generate image (completely free, no API key needed)
curl -sL -o output.jpg "https://image.pollinations.ai/prompt/description?width=1920&height=1080&nologo=true"

# Example: Tech-style AI concept image
curl -sL -o ai-concept.jpg "https://image.pollinations.ai/prompt/futuristic%20AI%20chip%20glowing%20blue%20on%20dark%20background%20minimalist%20tech%20style?width=1920&height=1080&nologo=true"

Prompt Best Practices

# Good prompt structure
{subject} + {style} + {color tone} + {composition} + {exclusions}

# Example
"Sam Altman speaking at conference, professional photo, warm lighting, 
 shallow depth of field, high resolution, no text no watermark"

# 16:9 cover image
"... 16:9 aspect ratio, horizontal composition, clean space on right for text overlay"

# Avoid garbled text
"... no text, no letters, no words, no watermarks"

Platform Comparison

Complete Workflow Example (All CLI, No Browser)

Scenario: Find image for "UK PM visits China" article

# Option A: Extract from Reuters (no browser)
# 1. Find article URL via web search or known URL
# 2. Extract image URL from HTML
IMG_URL=$(curl -sL "https://www.reuters.com/world/uk-pm-china-visit/" | \
  grep -oE 'https://www\.reuters\.com/resizer/v2/[^"]+\.jpg' | head -1)
# 3. Download with HD params
curl -sL -o starmer-xi.jpg "${IMG_URL}?width=3000&quality=100"

# Option B: Brave search (no browser)
IMG_URL=$(curl -s "https://api.search.brave.com/res/v1/images/search?q=Keir%20Starmer%20Xi%20Jinping%202025&count=1" \
  -H "X-Subscription-Token: $BRAVE_API_KEY" | jq -r '.results[0].properties.url')
curl -sL -o starmer-xi.jpg "$IMG_URL"

# Option C: AI generate (no browser)
curl -sL -o uk-china.jpg "https://image.pollinations.ai/prompt/British%20and%20Chinese%20flags%20diplomatic%20meeting%20minimalist?width=1920&height=1080&nologo=true"

# Verify result
file starmer-xi.jpg
ls -lh starmer-xi.jpg

Quick Reference Card

Choose Method by Scenario

Image Quality Checklist

• [ ] Resolution ≥ 1200px width
• [ ] File size > 50KB (smaller likely thumbnail)
• [ ] file command confirms image (not HTML)
• [ ] No watermarks
• [ ] No copyright issues (news images for commentary/citation)

Troubleshooting