Rendshot — Image Generation & Screenshot

Rendshot renders HTML/CSS to images, takes URL screenshots, and supports AI-powered template generation with platform presets.

Tool Selection

MCP tools available? Use them directly (preferred path):

Tool	Use when
generate_image	Render HTML, fill a template, or AI-generate from prompt
screenshot_url	Capture a webpage
list_templates	Browse community templates by platform/category
get_template	Get variable schema before filling a template
create_template	Save a design for reuse

No MCP tools? Fall back to CLI or API. See references/api-endpoints.md.

Core Workflows

1. AI Prompt to Image (fastest)

generate_image({
  prompt: "Coffee shop promotion card, warm brown tones, large title",
  platform: "xiaohongshu",    // auto-sets 1080x1440 + style guide
  locale: "zh"                // response language
})

Combine with template_id to reference an existing template's visual style:

generate_image({
  prompt: "Same layout but for a bakery",
  template_id: "tpl_abc123",  // style reference
  platform: "instagram_post"
})

2. Template to Image (deterministic)

// Step 1: Find a template
list_templates({ platform: "xiaohongshu", q: "product showcase" })

// Step 2: Check its variables
get_template({ template_id: "tpl_abc123" })
// Returns: title(text), subtitle(text), bg_image(image), accent_color(color)

// Step 3: Render with custom values
generate_image({
  template_id: "tpl_abc123",
  variables: {
    title: "Today's Special",
    subtitle: "Hand-roasted beans",
    bg_image: "https://example.com/beans.jpg",
    accent_color: "#8B4513"
  }
})

3. Raw HTML to Image

generate_image({
  html: "<div style='...'>Hello World</div>",
  width: 1080,
  height: 1080,
  format: "png",
  deviceScale: 2        // 2x for retina
})

4. Screenshot

screenshot_url({
  url: "https://example.com",
  width: 1280,
  height: 800,
  fullPage: true,
  format: "png"
})

5. Save and Reuse

After AI generates a design, save it as a template:

create_template({
  name: "Coffee Promo Card",
  html: "<returned html from AI>",
  variables: [
    { key: "title", type: "text", label: "Title", default: "Special Offer" },
    { key: "bg_image", type: "image", label: "Background", default: "https://..." }
  ],
  platform: "xiaohongshu",
  tags: ["coffee", "promotion"],
  visibility: "private"
})

Platform Presets

Platform	Default size	Typical use
xiaohongshu	1080x1440	Notes, product cards
xiaohongshu_wide	1080x810	Wide format notes
instagram_post	1080x1080	Feed posts
instagram_story	1080x1920	Stories, reels covers
twitter_card	1200x628	Link preview cards
twitter_post	1200x675	Image posts
linkedin_post	1200x627	Professional posts
youtube_thumb	1280x720	Video thumbnails
wechat_cover	900x383	Article covers
wechat_thumb	200x200	Article thumbnails
og_image	1200x630	Open Graph images
custom	1080x1080	Any custom size

Variable Types

Type	Example	Constraints
text	Title, subtitle	maxLength
image	Background, avatar	URL format
color	Accent, background	Hex format
number	Font size, opacity	min, max
select	Style variant	options: ["A","B"]

Key Rules

• html, template_id, and prompt are mutually exclusive (except prompt + template_id for style reference)
• Always call get_template before using a template to check its variable schema
• Use deviceScale: 2 for retina/high-DPI output
• Default format is png; use jpg + quality: 85 for smaller file size
• AI prompt mode returns HTML + variables; save good results with create_template
• For detailed tool parameters, see references/mcp-tools.md
• For API/CLI/SDK usage, see references/api-endpoints.md