@agentutility/mcp-synthforge

Generative media for agents that ship products.

Image, video, music, and voice generation with fast Venice models, pro creative models, and flagship Google/OpenAI image tiers. One API surface, USDC-settled, no SaaS account.

Pricing: pay-per-call in USDC on Base. No subscriptions, no API keys. See per-tool prices below.

Install — Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "agentutility-synthforge": {
      "command": "npx",
      "args": ["-y", "@agentutility/mcp-synthforge"],
      "env": { "X402_PRIVATE_KEY": "0xYOUR_PRIVATE_KEY_HEX" }
    }
  }
}

Restart Claude Desktop. 59 tools appear in the tool palette.

Install — Cursor

Add to .cursor/mcp.json:

{
  "mcpServers": {
    "agentutility-synthforge": {
      "command": "npx",
      "args": ["-y", "@agentutility/mcp-synthforge"],
      "env": { "X402_PRIVATE_KEY": "0x..." }
    }
  }
}

Funding

Send any amount of USDC on Base mainnet to the address derived from your X402_PRIVATE_KEY. The MCP server uses it to pay for tool calls automatically.

USDC on Base contract: 0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913

Tools (59)

| Tool | Description | |---|---| | ad-banner-image-expand | (0.15 USDC/call) Expands existing images into wide, square, or portrait ad banner placements. Same Bria Expand backend as image-expand. Use it as an ad banner image expand API, paid social canvas outpaint, or campaign creative extender. | | ad-creative-image-generate | (0.01 USDC/call) Generates ad creative images: SFW marketing visuals, concept images, and campaign thumbnails from prompts. Same backend as image-generate. Use it as a campaign visual generator or paid social creative API. | | app-icon-image-generate | (0.01 USDC/call) Generate app icon concepts, product marks, and launch visuals from a text prompt, ready for a mobile home screen or a SaaS dashboard. Same image-generate backend, restricted to SFW square outputs sized for icon use. Send a prompt describing the mark plus width, height, and a speed tier; get back a hosted image URL and the model used to render it. Use it as an app icon generator API, icon concept tool, or SaaS logo mark generator. | | app-store-screenshot-expand | (0.15 USDC/call) Expands mobile app screenshots into store, ad, and product-page layouts. Same Bria Expand backend as image-expand. Use it as an app store screenshot expand API, screenshot backdrop extender, or launch asset outpaint. | | background-remove | (0.08 USDC/call) Removes the background from a public image URL and returns the subject with alpha transparency. Optional crop_to_bbox. fal.ai imageutils/rembg. Use it as a background remove API, image background remover, or transparent PNG foreground cutout. | | background-remover | (0.08 USDC/call) Removes the background from product, portrait, and object images, returning alpha PNG foreground cutouts with optional crop-to-bounding-box behavior. Same remove-bg backend. Use it as a background remover API, to remove image background, or for transparent PNG cutouts. | | banner-image-expand | (0.15 USDC/call) Stretches an existing product or brand photo into a wide banner, hero, or ad canvas by generating new background content around it rather than cropping or distorting the original. Runs on the same image-expand outpainting backend: send image_url, target canvas_width and canvas_height, and an optional prompt describing the extended scene; get back the finished image_url plus its width and height. Use it as a banner image expand API, hero image outpainting tool, website header extender, or ad creative resizer for turning square product shots into wide-format assets. | | blog-header-image-generate | (0.01 USDC/call) Generates blog header images: SFW header images, newsletter visuals, and documentation illustrations from prompts. Same backend as image-generate. Use it as an article hero image or editorial illustration generator. | | book-cover-image-generate | (0.01 USDC/call) Generates SFW book cover art concepts, genre mood boards, and launch visuals from a prompt. Same image-generate backend. Use it as a book cover image generator API, ebook cover concept tool, or publishing creative source. | | brand-logo-concept-image | (0.01 USDC/call) Generates SFW logo concept boards, abstract marks, and brand direction visuals from prompts. Same fast Venice-backed backend as image-generate. Use it as a brand logo concept image API, for startup logo visual ideation, or as a mark concept generator. | | course-thumbnail-expand | (0.15 USDC/call) Expands existing screenshots, instructor photos, or diagrams into 16:9 course thumbnails. Same image-expand backend. Use it as a course thumbnail expand API, tutorial cover outpaint, or education hero image extender. | | course-thumbnail-image-generate | (0.01 USDC/call) Generates SFW course thumbnails, lesson covers, learning-path artwork, and academy launch visuals, without text rendering. Same fast image-generate backend. Use it as a course thumbnail image generator API, education cover art tool, or tutorial hero image source. | | ecommerce-lifestyle-image-generate | (0.01 USDC/call) Generates SFW ecommerce lifestyle scenes and ad visuals around product prompts. Same image-generate backend. Use it as an ecommerce lifestyle image generator API, product lifestyle creative tool, or marketplace campaign image source. | | email-header-image-expand | (0.15 USDC/call) Expands existing product, event, or editorial images into wide email-header canvases. Same image-expand backend. Use it as an email header image expand API, newsletter banner outpaint, or lifecycle email creative extender. | | event-poster-image-generate | (0.01 USDC/call) Generates SFW event artwork, launch posters, and campaign backgrounds without relying on stock art. Same fast image-generate backend. Use it as an event poster image generator API, conference visual tool, or meetup campaign image source. | | fashion-lookbook-image-generate | (0.01 USDC/call) Generates SFW fashion visuals: clothing concepts, seasonal lookbook scenes, and ecommerce fashion creative from prompts. Same image-generate backend. Use it as a fashion lookbook image generator API, apparel campaign visual tool, or style mood board source. | | flux-2-pro | (0.04 USDC/call) Generates SFW images with Venice's flux-2-pro model on a dedicated endpoint. Fixed-model route for buyers searching specifically for Flux 2 Pro / FLUX.2 Pro image generation; use image-generate-pro when you want a selectable premium model tier. Returns a permanent fal-hosted PNG URL. | | game-asset-concept-image | (0.01 USDC/call) Game asset concept image API / character prop environment concept generator. Same image-generate backend, exposed for agents drafting SFW game art concepts, item thumbnails, environment mood boards, and prototype visual assets. | | gpt-image | (0.30 USDC/call) Generates SFW images with Venice's gpt-image-2 model on a dedicated endpoint. Fixed-model route for OpenAI GPT Image 2 generation where instruction following and text rendering matter. Priced separately because GPT Image 2 has higher upstream cost than image-generate-ultra's $0.20 model set. Returns a permanent fal-hosted PNG URL. | | healthcare-campaign-image-generate | (0.01 USDC/call) Generates SFW healthcare, wellness, clinic, and patient-education campaign visuals, without medical claims. Same image-generate backend. Use it as a healthcare campaign image generator API, clinic marketing visual tool, or wellness program creative source. | | image-describe-api | (0.02 USDC/call) Describes images from public URLs, producing captions, accessibility alt text, short descriptions, and visual summaries. Same image-description backend. Use it as an image describe API, image captioning service, or alt text generator. | | image-edit | (0.02 USDC/call) Edits an image from a plain-text instruction, mask-free: describe the change and the model applies it to the whole image. Default model: nano-banana-pro. Returns a permanent fal-hosted PNG URL. Use it for instruction-based image editing, text-driven photo edits, nano-banana image editing, or GPT-image-2 edits. | | image-expand | (0.15 USDC/call) Extends a source photo onto a larger canvas, generating realistic new background and content beyond its original edges instead of just stretching or cropping it. Send image_url with canvas_width and canvas_height, optionally original_width/original_height and original_x/original_y to place the source image within the new canvas, plus an optional prompt, negative_prompt, and seed for control. Returns the expanded image_url with its final width, height, and seed, using a commercially licensed outpainting model. Use it as an image outpainting API, AI image expansion tool, or background extender for product photos and social crops. | | image-generate | (0.01 USDC/call) Generate an SFW text-to-image asset from a prompt, with optional dimensions, style tier, model controls, and seed. Returns a hosted PNG URL plus the resolved model, final dimensions, prompt, seed, and request timing. Choose fast for quick AI image generation, creative for stylized marketing art, anime for illustration, or sd35 for Stable Diffusion-style output. Use it as a text-to-image API, AI image generator, or prompt-to-PNG service for agents that need a hosted visual asset. | | image-generate-pro | (0.06 USDC/call) Premium text-to-image generation across margin-safe Venice models at a competitive $0.06/call. Tiers: 'balanced' (flux-2-pro, default), 'recraft' (recraft-v4), 'seedream' (seedream-v4), 'grok' (grok-imagine-image), and 'qwen' (qwen-image). Returns a permanent fal-hosted PNG URL. Use it for Flux 2 Pro, Recraft, Seedream, Qwen Image, or xAI Grok Imagine image generation. | | image-generate-ultra | (0.22 USDC/call) Flagship text-to-image generation for top proprietary Venice models that fit a $0.22 retail price. Tiers: 'nano-banana' (nano-banana-pro, default, Google Gemini Image 3), 'nano-banana-2' (cheaper Google variant), 'nano-banana-lite' (nano-banana-2-lite budget variant), and 'grok-sota' (grok-imagine-image-quality). Output is capped to 1024x1024 to keep wholesale within retail; for GPT Image 2 use the dedicated gpt-image endpoint. Returns a permanent fal-hosted PNG URL. | | image-inpaint | (0.02 USDC/call) Inpaints an image by replacing the masked region with content matching a text prompt. White pixels in the mask = region to inpaint. Default model: gpt-image-2. Returns a permanent fal-hosted PNG URL. Use it for mask-based image edits, object replacement, mask-driven face swap, or generative fill. | | image-to-video | (0.50 USDC/call) Animates a still image into video via fal.ai Bytedance Seedance. Optional prompt steers the motion (camera moves, subject motion). Returns a hosted MP4 URL plus fal request id and seed. Use it for image-to-video, Seedance image-to-video, motion-from-photo, or camera movement on a photo. | | linkedin-banner-expand | (0.15 USDC/call) Expands existing photos into wide LinkedIn banner and social-profile header canvases. Same Bria Expand backend as image-expand. Use it as a LinkedIn banner expand API, profile header outpainting tool, or brand banner image extender. | | marketplace-product-expand | (0.15 USDC/call) Expands marketplace product images, adapting product shots to marketplace, catalog, and ad aspect ratios. Same Bria Expand backend as image-expand. Use it as a product image expand API, ecommerce photo canvas extender, or product listing outpaint. | | mobile-story-image-expand | (0.15 USDC/call) Expands existing images into tall story, reel-cover, and short-form promotional canvases. Same image-expand backend. Use it as a mobile story image expand API, vertical social outpaint, or Instagram TikTok story extender. | | music-generate | (0.08 USDC/call) Generates music from a text prompt via Venice using ACE-Step 1.5. Optional lyrics input. Produces 60, 90, or 120 seconds of music; shorter requests round up to 60 seconds. Returns a permanent fal-hosted audio URL. Use it for AI music generation, text-to-music, generative songs, and instrumental or vocal music. | | nano-banana | (0.22 USDC/call) Generates SFW images with Venice's nano-banana-pro model on a dedicated endpoint. Fixed-model route for Nano Banana Pro / Google Gemini Image generation, strong visual reasoning, layout-aware creative, and high-end text-to-image output. Returns a permanent fal-hosted PNG URL. | | nano-banana-2-lite | (0.08 USDC/call) Generates SFW images with Venice's nano-banana-2-lite model on a dedicated low-cost endpoint. Fixed-model route for cheaper Nano Banana / Google image generation where buyers want the Nano Banana family at a lower price than nano-banana-pro. Returns a permanent fal-hosted PNG URL. | | newsletter-image-generate | (0.01 USDC/call) Generates SFW newsletter artwork, digest headers, and publication graphics. Same image-generate backend. Use it as a newsletter image generator API, email header illustration tool, or editorial campaign visual source. | | podcast-cover-expand | (0.15 USDC/call) Expands guest photos, product images, or show art into podcast-cover dimensions. Same image-expand backend. Use it as a podcast cover expand API, episode art outpaint, or square artwork canvas extender. | | podcast-cover-image-generate | (0.01 USDC/call) Generates SFW podcast cover art: square cover concepts, guest episode art, channel thumbnails, and audio-show campaign visuals. Same Venice-backed image-generate backend. Use it as a podcast cover image generator API, episode art tool, or show artwork concept source. | | presentation-hero-image-generate | (0.01 USDC/call) Generates SFW presentation visuals: deck openers, product narrative visuals, and board-slide hero images. Same image-generate backend. Use it as a presentation hero image generator API, pitch deck visual tool, or slide cover image source. | | product-image-generate | (0.01 USDC/call) Generates product images for ecommerce: SFW product concepts, studio-style mockups, and catalog creative from prompts. Same fast Venice-backed backend as image-generate. Use it as an ecommerce product creative or marketplace listing image generation API. | | product-photo-background-expand | (0.15 USDC/call) Expands tight product crops into marketplace, catalog, and lifestyle ad canvases by extending the background. Same image-expand backend. Use it as a product photo background expand API, ecommerce product outpaint, or studio backdrop extender. | | product-photo-expand | (0.15 USDC/call) Expands the canvas around an existing product shot for product-detail pages, marketplaces, and catalog creative. Same Bria Expand backend as image-expand. Use it as a product photo outpainting API, ecommerce image expand, or product hero image extender. | | qwen-image | (0.04 USDC/call) Generates SFW images with Venice's qwen-image model on a dedicated endpoint. Fixed-model route for Qwen Image generation, text-in-image attempts, bilingual prompt following, diagrams, posters, and structured visual concepts. Returns a permanent fal-hosted PNG URL. | | real-estate-ad-image-generate | (0.01 USDC/call) Generates SFW real estate ad visuals: property marketing imagery, neighborhood concept art, open-house ads, and listing campaign backgrounds. Same image-generate backend. Use it as a real estate ad image generator API, property campaign visual tool, or listing promotion creative source. | | real-estate-hero-expand | (0.15 USDC/call) Expands room, exterior, and neighborhood photos into wide property-page and ad hero canvases. Same image-expand backend. Use it as a real estate hero image expand API, listing banner outpaint, or property website image extender. | | real-estate-photo-expand | (0.15 USDC/call) Extends an existing room or exterior photo into a wider hero crop or vertical social version for listing creatives. Same backend as image-expand. Use it as a real estate photo outpainting API, property listing image expand, or room photo canvas extender. | | real-estate-render-generate | (0.01 USDC/call) Generates SFW real estate visuals: listing concepts, room mood boards, exterior concepts, and property campaign imagery. Same image-generate backend. Use it as a real estate render generator API, property marketing image tool, or interior concept visual source. | | recraft | (0.06 USDC/call) Generates SFW design and illustration images with Venice's recraft-v4 model on a dedicated endpoint. Fixed-model route for Recraft image generation, brand visuals, icon concepts, design mockups, and text-aware illustration. Returns a permanent fal-hosted PNG URL. | | remove-background | (0.08 USDC/call) Removes the background from an image URL and returns the subject as a PNG with alpha channel. Optional crop_to_bbox. fal.ai imageutils/rembg. Use it as a background remove API or transparent PNG cutout. | | remove-bg | (0.08 USDC/call) Removes the background from an image and returns a foreground cutout with alpha channel. Optional crop_to_bbox. fal.ai imageutils/rembg. Use it as an AI background remover, background eraser, or transparent PNG cutout tool. | | restaurant-menu-image-generate | (0.01 USDC/call) Generates SFW restaurant imagery: dish concepts, menu hero visuals, and local restaurant ad creative. Same image-generate backend. Use it as a restaurant menu image generator API, food concept visual tool, or hospitality creative source. | | seedance-video | (0.50 USDC/call) Generates AI video with fal.ai Bytedance Seedance. Duration, aspect ratio, and resolution are configurable. Same backend as text-to-video under a model-named slug for direct discovery by agents searching for Seedance. Use it for Seedance video generation, text-to-video AI, and cinematic AI clips. | | seedream | (0.06 USDC/call) Generates SFW images with Venice's seedream-v4 model on a dedicated endpoint. Fixed-model route for Seedream image generation, photoreal creative, product concepts, and fast high-quality campaign visuals. Returns a permanent fal-hosted PNG URL. | | social-crop-expand | (0.15 USDC/call) Adapts one image into square, portrait, landscape, and banner canvases. Same backend as image-expand, for agents reformatting creative per platform. Use it as a social crop expand API, Instagram TikTok LinkedIn crop extender, or image aspect-ratio outpainting tool. | | social-image-generate | (0.01 USDC/call) Generates safe-for-work images sized for social posts, in square, portrait, and landscape formats, from a text prompt. Send prompt plus optional width, height, and tier (fast, creative, anime, or sd35, each mapped to a Venice image model); the endpoint always renders with safe_mode on and returns image_url, the resolved model, and final width and height. Same backend as image-generate, so pricing and model behavior match that endpoint's cheap tier. Use it as a social media image generator, post visual generator, or LinkedIn, X, and Instagram creative API for marketing teams and agents that need on-brand visuals without touching a design tool. | | sound-effect-generate | (0.01 USDC/call) Generates sound effects from a text prompt via Venice using elevenlabs-sound-effects-v2. Duration 0.5-22 seconds. Returns a permanent fal-hosted audio URL (or Venice-hosted when applicable). Use it as a text-to-SFX tool, Foley generator, ElevenLabs sound effects endpoint, or ambient audio synth. | | text-to-speech | (0.05 USDC/call) Converts text to speech with 30+ voices and 5 audio formats. Morpheus primary for Kokoro, Venice fallback and alternate TTS models (xAI / ElevenLabs / Orpheus / MiniMax / Gemini), with fal.ai storage for hosted audio URLs. Use it as a TTS API or voice generator. | | text-to-video | (0.50 USDC/call) Generates video from a text prompt via fal.ai Bytedance Seedance. Duration, aspect ratio, and resolution are configurable. Returns a hosted MP4 URL plus fal request id and seed. Use it for AI video, Seedance generation, generative video, and cinematic clips from a prompt. | | voice | (0.05 USDC/call) Turn written text into spoken audio you can drop straight into a player or app. Send text (up to 4,000 characters) with an optional voice, model, speed, and output format of mp3, wav, opus, aac, or flac, and get back a hosted audio_url plus file size and content type, no base64 handling required. Runs on Venice's text-to-speech models with 30+ selectable voices and adjustable playback speed from 0.25x to 4x. Use it as a text-to-speech API, voice synthesis tool, or narration generator for voiceovers, IVR prompts, or accessibility read-aloud features. | | youtube-thumbnail-expand | (0.15 USDC/call) Expands source images into YouTube thumbnail and video-cover canvases. Same Bria Expand backend as image-expand. Use it as a YouTube thumbnail expand API, video thumbnail outpaint, or 16:9 image extender. |

How it works

1. Agent calls a tool (e.g. ad-banner-image-expand).
2. MCP server POSTs to https://x402.agentutility.ai/ad-banner-image-expand.
3. The endpoint responds HTTP 402 with payment instructions.
4. The MCP server signs an EIP-3009 USDC transfer authorization with X402_PRIVATE_KEY and retries.
5. CDP facilitator settles on Base.
6. The endpoint returns the actual response.

The agent never sees the payment flow — it just gets the result.

Links

• Cluster overview: https://agentutility.ai/synthforge/
• All MCP packages: https://mcp.agentutility.ai/
• Source: https://github.com/rooz21/x402/tree/main/packages/mcp-synthforge

Version: 0.8.8 · License: MIT