MCP/Tools/Images/generate_image
generate_image

Generate an ad creative image

Medium cost

Generate a single (or multi-aspect-ratio batch) ad creative from a director-style prompt, with optional brand kit + logo composite.

How it works

The workhorse image tool. Pass a prompt, optionally hand it a brandId - the tool weaves your brand colors, voice, aesthetic, and saved logo into the render. Supports every Claivra-supported aspect ratio plus ad-spec presets like 1.91:1 (auto-resized to the exact 1200×628 Google Display Network spec). Pass an array via aspectRatios and the tool produces one image per ratio in a single bounded-concurrency batch.

Inputs

Name
Type
Required
Description
prompt
string
required
Director-style brief describing the image to render.
model
'claivra-v1' | 'claivra-v2' | 'claivra-v3' | 'claivra-v4' | 'claivra-v5'
optional
v1 = fast photorealism (1 credit, 1K/2K/4K). v2 = dense posters/UI & text (4 credits, 1K/2K/4K). v3 = best-in-class typography/text (3 credits - no logo composite, native res). v4 = fast vivid photoreal (2 credits - no logo composite, native res). v5 = premium, crisp text (4-7 credits, 1K/2K/4K). Default v1.
resolution
'1K' | '2K' | '4K'
optional
Output resolution (default 1K). Honored by v1, v2, v5 - higher = more credits. v3/v4 render native (ignored).
aspectRatio
string
optional
Single ratio (1:1 default). Use aspectRatios for batch.
aspectRatios
string[]
optional
Multi-ratio batch - renders one image per ratio in parallel.
brandId
string
optional
Brand kit ID - wires colors + logo + voice into the prompt automatically.
logo
{ data: base64, mimeType: string }
optional
Override the brand's saved logo for this generation only.
referenceImages
Array<{ data, mimeType }>
optional
Style references the model anchors to.
featureBrandIds
string[]
optional
Other brand ids whose REAL logos to pull into the image (e.g. a client logo in an agency post) - fetched server-side so the model reproduces the actual logo, not an invented one.
cta
string
optional
CTA button text to overlay on the image.
tone
string
optional
Tone descriptor woven into the prompt.

Returns

{ imageId, imageUrl, mimeType } - or an array of refs when aspectRatios has 2+ entries.

Examples

Square ad with brand kit applied

Agent generates a 1:1 hero ad for a client's automotive brand. Brand colors, fonts, and logo are baked in via brandId.

{
  "prompt": "Hero shot of a green Rivian R1S parked in a sleek modern showroom with dramatic accent lighting. Bold headline overlay: 'EVs that mean business.' CTA pill bottom-center.",
  "model": "claivra-v1",
  "aspectRatio": "1:1",
  "brandId": "aBmbZETe9gUZhwXHCE02",
  "cta": "Get Your Quote"
}

Square + Google Ad in one call

Generate the same concept at both 1:1 (Instagram feed) and 1.91:1 (Google Display Network) - resized server-side to the exact 1200×628 spec.

{
  "prompt": "Minimalist studio shot of a Tesla Model S with bold typography overlay 'Drive into the future.'",
  "aspectRatios": [
    "1:1",
    "1.91:1"
  ],
  "brandId": "ectL5JsJKz55ItiFzEkB"
}

Related tools

edit_image

Edit an existing image with an instruction

Apply a natural-language instruction to an existing image - swap headlines, change colors, remove text, replace backgrounds - without regenerating the whole composition.

generate_variations

Generate variations of a reference image

Re-skin / copy any ad (the 'AI Ad Copier'): upload a reference creative (+ optional logo), pick a Logo mode, add an offer / filters / resolution, and get N on-brand variant concepts that keep the structural framing.

upload_image

Upload an image into your Claivra account

Push a base64 image you already have on disk (cropped, resized, brand-overlayed) into your Claivra media library - usable in create_share, revise_share, edit_image.

All toolsGet API key