Product photos
Use QuokkaPix to prepare these local images for Shopify: white background, square product canvas, WebP output, max 300 KB, then save the ZIP and result manifest.
Agent integration
Connect local or remote AI agents to QuokkaPix Image and Video through one npm adapter and a browser-local execution bridge.
Version 0.7.0 exposes one exact Image and Video quote/token flow through local stdio and the OAuth-protected remote MCP with a paired bridge. Use process_video with an official recipe or process_video_with_settings with direct settings. Video runs in local Chromium and never relays media bytes through the control plane.
{
"recipeId": "tiktok_vertical",
"inputFile": "/absolute/path/source.mp4",
"outputDir": "/absolute/path/output"
}
Claude web and other cloud MCP clients can use the OAuth-protected remote endpoint while the npm adapter executes Image and Video operations on this computer.
npx quokkapix-mcp bridge --input-root /path/to/input --output-root /path/to/output
Remote MCP: https://quokkapix.com/mcp
The mcp:tools scope permits read tools. Processing and staged unlocks also require bridge:execute. The --reset option revokes the previous device before pairing a replacement.
The control plane relays settings, relative file names, status and result metadata. It has no media upload endpoint and does not relay source or output image, video or audio bytes.
Use this page when you want a desktop AI agent to process local image files with QuokkaPix. The MCP runner is a local stdio adapter: the MCP client starts quokkapix-mcp, the adapter opens QuokkaPix in a local browser, uploads files through the browser file input, applies a recipe or direct settings payload, downloads the result and writes quokkapix-result.json.
This is not a hosted image-processing API. Source images stay in the local browser workflow instead of being uploaded to a QuokkaPix processing server.
Add this server entry in Claude Desktop MCP settings, then restart Claude Desktop. After restart, confirm that tools such as list_recipes, process_images and process_with_settings are visible.
{
"mcpServers": {
"quokkapix": {
"command": "npx",
"args": [
"-y",
"quokkapix-mcp"
],
"env": {
"QUOKKAPIX_APP_URL": "https://quokkapix.com/#agent=1"
}
}
}
}
Use Cursor MCP settings and add the same server entry. Cursor starts the adapter over stdio and can call the same tools as Claude Desktop.
{
"mcpServers": {
"quokkapix": {
"command": "npx",
"args": [
"-y",
"quokkapix-mcp"
],
"env": {
"QUOKKAPIX_APP_URL": "https://quokkapix.com/#agent=1"
}
}
}
}
If a Windows MCP client cannot resolve npx directly, start it through cmd /c. Use this only when the simple npx command is not detected by the client.
{
"mcpServers": {
"quokkapix": {
"command": "cmd",
"args": [
"/c",
"npx",
"-y",
"quokkapix-mcp"
],
"env": {
"QUOKKAPIX_APP_URL": "https://quokkapix.com/#agent=1"
}
}
}
}
Any MCP-compatible local client that can start a stdio server can use the same package. This includes desktop clients, editor agents, LM Studio or Ollama wrappers that expose MCP tools to a local model.
The important rule is the same everywhere: pass real local image file paths to the MCP tool call, not to the public website URL. QuokkaPix receives files only through the local browser file input opened by the runner.
Use QuokkaPix to prepare these local images for Shopify: white background, square product canvas, WebP output, max 300 KB, then save the ZIP and result manifest.
Use QuokkaPix to convert this folder of JPG and PNG images to WebP for the website, keep reasonable visual quality, and report files that fail QA.
Use QuokkaPix to remove EXIF, GPS, camera and software metadata from these images and return the cleaned ZIP plus the manifest.
Use list_recipes first, choose a recipe id, then call process_images. Official recipes cover common marketplace, social, compression, metadata, PDF, favicon and watermark workflows.
{
"recipeId": "shopify_product_pack",
"inputFiles": [
"/Users/me/products/photo-01.jpg",
"/Users/me/products/photo-02.jpg"
],
"outputDir": "/Users/me/quokkapix-output/shopify"
}
Use process_with_settings when the agent needs the same settings surface as window.QuokkaPixAgent.applySettings instead of a named recipe.
{
"settings": {
"mode": "batch",
"tool": "compress",
"settings": {
"compress": {
"format": "webp",
"quality": 0.82
}
}
},
"inputFiles": [
"/Users/me/site/hero.jpg",
"/Users/me/site/card.jpg"
],
"outputDir": "/Users/me/quokkapix-output/webp"
}
| Check | Why it matters |
|---|---|
Output file, ZIP or PDF exists in outputDir | Confirms that the browser download completed and the adapter saved the deliverable. |
quokkapix-result.json has status: done | Confirms that QuokkaPix reached a terminal successful state instead of only opening the editor. |
outputs[] contains source and output metadata | Lets the agent report dimensions, formats, file sizes, warnings and file names. |
qa.ok and qa.checks[] when present | Shows whether a marketplace or social recipe passed known rule checks and what must be fixed. |
Image: 1-5 actual files are free, 6-25 cost 0.01 USDC, and 26-50 cost 0.02; PDF is free through 20 pages, then 0.01 through 50 and 0.02 through 300. A multi-step Image scenario costs 0.02. Video editing is free through 5 minutes, then 0.01 / 0.02 / 0.03; transcription is free through 5 minutes, then 0.02 / 0.04 / 0.06 through 45 minutes.
Upload first and obtain the exact local quote. Pay its product endpoint, then retry the unchanged run with unlockToken. The browser consumes it only immediately before processing; WebMCP and Bridge add no fee.
Restart the MCP client and confirm that Node.js can run npx -y quokkapix-mcp from a terminal.
Check that the local environment allows Playwright/Chromium to start. The adapter needs a browser because QuokkaPix is browser-only.
Check outputDir, browser memory, the terminal status and errorCode in quokkapix-result.json.
Fetch current payment options, obtain a valid x402 unlock token and set it before starting a batch above the free small-batch limit.
Use the official QuokkaPix URL or a trusted local development URL. Do not point the runner at an arbitrary third-party appUrl, because the adapter uploads local files into the browser page it opens. The public website does not receive local file paths, and the adapter does not turn QuokkaPix into a public server-side image API.
No. The adapter opens QuokkaPix in a local browser and selects files through the browser input. Image and Video workflows run in the browser runtime.
Yes, for coordination only. The remote MCP endpoint requires a paired local quokkapix-mcp bridge; Image and Video processing and all media bytes remain on the user's device.
Image: 1-5 actual files are free, 6-25 cost 0.01 USDC, and 26-50 cost 0.02; PDF is free through 20 pages, then 0.01 through 50 and 0.02 through 300. A multi-step Image scenario costs 0.02. Video editing is free through 5 minutes, then 0.01 / 0.02 / 0.03; transcription is free through 5 minutes, then 0.02 / 0.04 / 0.06 through 45 minutes.
Use the runner when an agent needs to process local image files without sending the source files to a QuokkaPix server. The adapter starts a local browser session, uploads files through the normal file input, applies an official recipe or direct settings, then saves the downloaded output and manifest to the output directory.
The runner should open the official QuokkaPix app URL or a trusted local development URL. It does not turn QuokkaPix into a public server-side processing API; files are still selected through the browser and processed on the machine running the browser.
list_recipesget_recipevalidate_recipelist_rule_profilesget_rule_profilevalidate_result_manifestprocess_imagesprocess_with_settingslist_video_recipesget_video_recipevalidate_video_recipelist_video_rule_profilesget_video_rule_profilevalidate_video_result_manifestprocess_videoprocess_video_with_settingsget_payment_optionsexplain_payment_flowverify_unlock_tokenshopify_product_packPrepare product photos for Shopify listings.
batch · zip
amazon_white_background_packPrepare marketplace product photos with white background, square canvas and compressed output.
batch · zip
google_merchant_packPrepare consistent product images for Google Merchant feeds.
batch · zip
etsy_product_batchPrepare Etsy product images with supported JPG output and a sourced Etsy QA profile.
batch · zip
ebay_listing_photo_batchResize listing photos toward eBay's sourced long-side guidance and export a local ZIP.
batch · zip
walmart_product_main_batchPrepare white-background product images using sourced Walmart image guidance.
batch · zip
tiktok_shop_product_batchPrepare TikTok Shop product images against sourced image size and content checks.
batch · zip
temu_product_main_batchPrepare Temu-style product images using a secondary sourced profile because public official specs are limited.
batch · zip
shopee_product_batchPrepare Shopee product images against sourced minimum dimensions from Shopee seller education.
batch · zip
mercado_libre_accessories_batchPrepare Mercado Libre accessory photos using the sourced 1200 x 1540 px category profile.
batch · zip
allegro_listing_image_batchPrepare Allegro listing images against sourced long-side, max-size and format rules.
batch · zip
newegg_product_image_batchPrepare Newegg product images with white background, minimum dimensions and file-size QA.
batch · zip
meta_catalog_product_batchPrepare Meta catalog product images with sourced JPEG/PNG, 500 px minimum and 8 MB size checks.
batch · zip
flipkart_product_image_batchPrepare clean white-background Flipkart product images using public seller photography guidance.
batch · zip
shein_product_square_batchPrepare SHEIN square product images using a secondary sourced 900 px profile.
batch · zip
otto_product_image_batchPrepare OTTO Market product images against official JPG/PNG, RGB and 500 x 1000 px minimum image rules.
batch · zip
trendyol_product_image_batchPrepare Trendyol product images at the official 1200 x 1800 px product image size.
batch · zip
snapchat_ad_image_batchPrepare Snapchat 9:16 single image ads using sourced 720 x 1280 requirements.
batch · zip
website_webp_compressCompress website images to WebP locally without uploading source files.
batch · zip
webp_compress_batchConvert and compress a batch of images to WebP for general web use.
batch · zip
white_background_shadow_batchRemove or replace backgrounds, add a white product canvas, soft shadow and compressed WebP output.
batch · zip
metadata_clean_batchRemove EXIF, GPS, camera and software metadata from images.
batch · zip
single_webp_compressCompress one image to WebP locally without using a paid batch unlock.
single · image
single_background_removeRemove the background from one image with transparent PNG/WebP output.
single · image
single_white_backgroundCreate one white-background product image with clean edges and compressed WebP output.
single · image
single_metadata_cleanRemove EXIF, GPS, camera and software metadata from one image.
single · image
single_watermarkApply a visible text watermark to one image.
single · image
images_to_pdf_batchMerge selected images or scans into one local PDF without uploading the files.
batch · pdf
social_pack_singleCreate common social media output sizes from one image.
single · zip
profile_avatar_packGenerate common square avatar sizes from one source image.
single · zip
watermark_product_batchApply a repeatable tiled watermark to a batch of product images.
batch · zip
favicon_app_icon_packGenerate favicon and app icon sizes from a logo image.
single · zip
{
"settings": {
"mode": "single",
"tool": "compress",
"settings": {
"compress": {
"format": "webp",
"quality": 0.82
}
}
},
"inputFiles": [
"/Users/me/photo.jpg"
],
"outputDir": "/Users/me/quokkapix-output"
}
Images, videos and audio still enter through local browser inputs. QuokkaPix does not expose a public server-side media-processing API.