Playground AI MCP. Generate, edit, and scale visuals with code.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Playground AI connects your agent directly to advanced image computation clusters. It lets you generate new images using text prompts, surgically edit existing photos with masks, radically extend borders, and scale low-res assets up to 4x—all without opening a visual editor.
What your AI agents can do
Generate image
Creates new images from a text prompt, allowing you to specify model type and dimensions.
Generate with controlnet
Generates images using ControlNet guidance, matching the structure or pose of a reference image.
Get generation
Retrieves detailed information about a previous Playground AI generation ID.
Creates new visual assets entirely from a text prompt, allowing you to control size and model variants.
Allows targeted repairs or expansions on existing images by passing masks and directional prompts.
Automatically isolates a subject from a photo, returning the result as a clean, transparent PNG file.
Increases the resolution and detail of existing images up to 4x or expands them beyond their original frame.
Generates new images while forcing the output structure (like depth or pose) to match a provided reference image.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
Playground AI MCP Server: 10 Tools for Visual Pipelines
Access ten tools to handle the entire lifecycle of digital media: generation, structural editing, scaling, and refinement.
019d75f7generate image
Creates new images from a text prompt, allowing you to specify model type and dimensions.
019d75f7generate with controlnet
Generates images using ControlNet guidance, matching the structure or pose of a reference image.
019d75f7get generation
Retrieves detailed information about a previous Playground AI generation ID.
019d75f7inpaint image
Fills in specific areas of an image using a mask and a text prompt to guide the edit.
019d75f7list generations
Lists recent generation IDs, prompts, and timestamps for tracking purposes.
019d75f7list models
Returns a list of available Playground AI models and their specific capabilities.
019d75f7outpaint image
Extends an image beyond its current borders, generating new content based on direction and prompt.
019d75f7remove background
Strips the background from a given image, returning a transparent PNG file of the subject.
019d75f7transform image
Changes an existing image using a text prompt and a strength control value (0 to 1).
019d75f7upscale image
Increases the resolution and detail of a provided image by a factor of 2 or 4.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Playground AI, then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
Generating New Visual Assets
You’ll generate brand new images using text prompts with generate_image. You just pass a prompt and specify dimensions, letting you control the size of your output. For more structured results, you can use generate_with_controlnet to force the image's pose or depth map to match a specific reference photo. If you need to see what models are available before starting, run list_models to get a full rundown of Playground AI’s capabilities.
Targeted Editing and Expansion
When you gotta fix an existing picture, you've got surgical options. Use inpaint_image when you need to fill in specific spots—you pass it a mask showing the area and a prompt telling it what to draw there. If you wanna expand the edges of a photo, outpaint_image lets you grow the borders using directional prompts so it generates new content that matches the existing scene.
You can also change an entire image's look dramatically with transform_image, feeding it a text prompt and a strength control value (from 0 to 1) to dictate how much the original picture gets modified.
Isolation and Enhancement
If you just need the subject, use remove_background to automatically strip out the background from any photo. The output’s always a clean, transparent PNG file of what's left. For blurry or low-res shots, run upscale_image. This tool increases both the resolution and the detail in an image by a factor of 2 or 4.
You can also get more specific data about past work using list_generations, which tracks previous generation IDs, prompts, and timestamps. To check the details on any single piece of art you made before, just use get_generation with the ID.
How Playground AI MCP Works
- 1 First, you append the Playground integration and provide your API key within the MCP setup.
- 2 Next, you instruct your agent to perform an action—for example, running
remove_backgroundon a photo URL. - 3 Finally, the server executes the tool call, returns the resulting file URL (like a transparent PNG), which your code then processes.
The bottom line is you run complex visual pipelines by calling simple, defined tools from within your code base.
Who Is Playground AI MCP For?
Anyone dealing with digital assets—concept artists, web developers, and creative directors. You're the person who gets frustrated when a marketing asset is slightly the wrong size or needs minor cleanup before deployment. This lets you fix those visuals programmatically without opening Photoshop.
Needs to generate perfectly sized graphical placeholders for UI components on demand, instead of manually creating and adjusting assets in an image editor.
Uses the agent's remove_background tool repeatedly to strip subjects from reference photos quickly, automating repetitive asset prep work.
Runs quick iterative tests—like applying an instantaneous upscale_image adjustment or extending a scene using outpaint_image—while discussing copy with the team.
What Changes When You Connect
- Stop opening a visual editor for basic tasks. You can run
remove_backgrounddirectly from your script to get transparent PNGs instantly, feeding that asset straight into the next step of your pipeline. - Need to change an old graphic? Use
transform_image. Instead of manually tweaking layers, you pass a prompt and a strength value (0-1) to mutate the image programmatically. - When your initial concept is solid but too small, run
upscale_imagewith factor 4. You get high-res results without losing detail, making it perfect for web assets. - Don't rely on luck for composition. Use
generate_with_controlnetto ensure new images adhere strictly to a reference pose or depth map, giving you predictable output every time. - Fixing mistakes is easy now. If an asset needs a detail added—say, a button on a character’s jacket—use
inpaint_imagewith a mask and prompt instead of doing tedious manual photo-editing.
Real-World Use Cases
Building UI Placeholders
You're building a landing page mockup and need 10 different hero images, but you don't want to open Playground AI. You call generate_image ten times in a loop, passing slight variations in the prompt for each placeholder. Your agent returns all necessary assets with specific dimensions (e.g., 800x600), letting your code build the page structure automatically.
Automating Character Asset Prep
A concept artist has a folder of reference photos for characters that need to be used in different backgrounds. Instead of cropping and masking them manually, they write a script that iterates through the folder, running remove_background on every image. The agent collects all clean PNG assets into a single output directory.
Expanding Scene Backgrounds
Your team created a great piece of art, but it's too narrow for the new billboard size. You pass the original image and call outpaint_image with a prompt like 'continuation of the forest floor'. The AI expands the scene on the left and right while maintaining consistent lighting and style.
Upscaling Batch Graphics
You receive 50 small, low-resolution icons from a vendor. Instead of running them through an external upscaler GUI, you feed the URLs into your agent which calls upscale_image with factor 4 on all 50. You get 50 high-res assets back in one go.
The Tradeoffs
Manually scaling images
A developer sees a low-res icon and tries to resize it using standard image manipulation libraries, resulting in blurry, pixelated artifacts that lose the original artwork's detail.
→
Use upscale_image with factor 4. This tool uses sophisticated AI interpolation to enhance resolution and detail rather than just stretching pixels.
Trying to change style in a GUI
A designer wants to take an existing product photo but needs it rendered as if it were painted by Van Gogh. They spend time searching for filters or manual effects controls.
→
Use transform_image. Pass the original image URL, your prompt ('Van Gogh oil painting style'), and set a high strength value (e.g., 0.8) to mutate the existing photo with text-guided art.
Forgetting masks for edits
You want to change the shirt color on a photo, but you just pass the image and prompt 'change shirt to blue.' The AI changes everything—the background, skin tone, everything—because it doesn't know exactly where to focus.
→
Always use inpaint_image. Pass the original image, your prompt ('blue shirt'), AND a precise mask image URL that only covers the area you actually want changed.
When It Fits, When It Doesn't
Use this server if your workflow requires chained visual tasks: Generation -> Modification -> Refinement. It's perfect for pipelines where an output from Tool A becomes the input for Tool B (e.g., remove_background -> outpaint_image).
Don't use it if you just need simple file storage or basic asset retrieval—a standard object store is fine. Also, don't rely on this for complex 3D visualization; the tools are focused purely on 2D image plane manipulation.
If your core problem is 'I need to generate a new visual based on an existing concept,' then you need generate_with_controlnet. If your problem is simply 'I need more pixels,' use upscale_image or outpaint_image. Always check if the tool requires a mask; if so, don't skip it.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Playground AI. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 10 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Manually preparing assets for web deployment is a nightmare.
Right now, every time you launch a new campaign, someone has to open the original image file. They manually crop it into different aspect ratios (1:1 for Instagram, 16:9 for hero banners). Then they find the background removal tool, click through the masking interface, and export separate PNGs. If you need five sizes, that's at least three different tools and two hours of clicking.
With this MCP server, your agent handles it all. You call a sequence of tools—say, `remove_background` first to clean up the subject. Then, you pass the resulting transparent image URL into a tool chain that automatically runs `upscale_image` and finally saves multiple cropped versions based on predefined dimensions. The asset is ready in your code.
Playground AI MCP Server: Run advanced visual pipelines from code.
The old way meant jumping between a design tool, an image editor, and then the background removal service. It was copy-paste hell with manual quality checks at every stage. If you changed one thing—like adding a subtle glow to the subject—you had to re-run the entire chain.
Now, it's a single sequence of API calls. You modify the image using `transform_image`, and that change instantly propagates through any subsequent tool call, like running `outpaint_image`. It's fully programmatic.
Common Questions About Playground AI MCP
How do I use generate_image for a specific size? +
You include the width and height (which must be multiples of 64) directly in your prompt parameters. For example, you'd specify '1024x1024' to get an image that large.
Can I use inpaint_image if I don't know what the mask is? +
No, inpaint_image requires three things: the prompt, the source image URL, and a separate mask image URL. The mask tells the AI exactly which pixels are editable.
What's the difference between outpaint_image and generate_image? +
generate_image makes an entirely new picture from scratch based on text. outpaint_image, however, uses your existing picture as a guide and expands its borders while maintaining continuity.
How does upscale_image improve quality? +
upscale_image doesn't just stretch the pixels; it runs an AI algorithm that intelligently interpolates missing data, adding detail and enhancing resolution up to 4x.
How do I use the `list_models` tool to check available Playground AI models? +
Run list_models to pull a current manifest of all active models. This is key for ensuring your generation calls don't fail because you are referencing an outdated or unsupported model name.
What does the strength parameter control when I use `transform_image`? +
The 'strength' value (0 to 1) dictates how much the AI changes the original pixels. Use a low number to keep structural integrity, or ramp up the strength for radical visual mutation.
If I run `generate_image` and need the result later, how do I use `get_generation`? +
Simply pass the unique generation ID to get_generation. This retrieves all metadata—images, prompt history, model info—allowing you to track job status even if the URL isn't immediately available.
What structural guidance does ControlNet provide when I use `generate_with_controlnet`? +
ControlNet forces the generated image to adhere strictly to a specific input map. You can guide depth, pose skeletons, or edge detection (canny), guaranteeing anatomical and spatial accuracy that simple prompting cannot achieve.
Can my AI automatically remove a background and embed the image into my project? +
Yes! The agent triggers the remove_background method on any public image URL. Playground internally isolates the core subject, dropping the background layers, and replies with a secure pointer URL. Your agent can instantly fetch the response string and code an `` tag placing the clean asset directly into the UI mapping you're developing.
How exactly does `outpaint_image` expand the canvas? +
The tool passes the target image alongside the literal specific expansion vectors ('up', 'down', 'left', 'right') and a guiding context text. The diffusion nodes generate purely new mathematical tensor states radiating outward, predicting the lighting, shadows, and environment strictly following your prompt rules without stretching existing pixels.
Can I enforce specific structural references using ControlNet constraints? +
Yes. Instead of standard generation, invoke generate_with_controlnet. You upload a scribble or an edges-only mapping layout as your source. By defining the explicit type ('canny', 'depth', 'pose', 'scribble'), the AI strictly anchors its conceptual render to those physical guiding constraints rather than randomizing spatial architecture.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
Together AI
Generate code, evaluate embeddings, and deploy open-source LLMs instantly from your local agent via Together AI's infrastructure.
Vectara
Empower your agent with Vectara's RAG capabilities. Search corpora natively, execute grounded chats, and manage indexed datasets easily.
Spellbook Legal AI
AI-powered contract drafting and review — analyze contracts, draft clauses, detect risks, and compare against 2,000+ market precedents via Spellbook.
You might also like
Fathom Alternative
Privacy-first website analytics — track visitors, monitor real-time traffic, and manage sites and events directly from your AI agent.
Clearscope
Optimize content via Clearscope — create SEO reports, grade drafts using NLP, perform keyword research, and audit SERP competitors directly from any AI agent.
Transistor.fm
Host your podcast with analytics, multiple show support, and distribution to Apple Podcasts, Spotify, and every major platform.