ridvay-mcp
This server connects AI assistants to Ridvay Studio, enabling creation, editing, and management of graphic designs (posters, flyers, social media posts) directly from chat.
Generate designs from a text brief: Describe what you want and Ridvay's AI engine creates the design automatically (~20–40 seconds, uses generation credits). Supports square, story, portrait, landscape, A4, slide, or custom canvas sizes.
Compose your own designs: The AI assistant builds the layout using Ridvay's design IR format and saves it to Ridvay for rendering — no Ridvay-side AI generation consumed, giving full creative control.
Learn the design format: Retrieve the Ridvay design IR specification (element types, fonts, backgrounds, worked examples) to compose designs from scratch.
Edit existing designs: Apply natural-language edits to a previously created design (e.g., "make the headline red", "switch to a dark theme") using a design ID.
Check rendering status: Poll whether a design's AI-generated images have finished rendering and retrieve updated view/share/edit links.
Sharing control: Publish designs as unlisted public share links or keep them private to the account.
Brand Kit integration: Optionally apply the account's brand colors, fonts, and logo when generating or refining designs.
Allows GitHub Copilot (VS Code agent mode) to create and edit Ridvay Studio posters, flyers, and social designs directly from chat.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@ridvay-mcp Create a poster for our weekend flash sale – 30% off everything"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.

ridvay-mcp
MCP (Model Context Protocol) server that lets AI assistants — Claude Code, Claude Desktop, GitHub Copilot (VS Code agent mode), Cursor, and any other MCP client — create and edit Ridvay Studio posters, flyers, and social designs straight from a chat conversation.
Every design comes back as a working share link (ridvay.com/d/…) plus links to open and
edit it in the Studio editor.
Hosted endpoint (no install)
The same server runs at https://mcp.ridvay.com (streamable HTTP). Two ways to authenticate:
URL with your key — for clients that add connectors by URL and cannot set headers (claude.ai custom connectors, ChatGPT connectors):
https://mcp.ridvay.com/mcp/sk-ridvay-…. Treat that URL like the key itself; revoke it any time at ridvay.com/user/api-keys.Bearer header — for clients that support headers:
https://mcp.ridvay.com/mcpwithAuthorization: Bearer sk-ridvay-…. Claude Code:claude mcp add --transport http ridvay https://mcp.ridvay.com/mcp --header "Authorization: Bearer sk-ridvay-…".Try it without an account —
https://mcp.ridvay.com/mcp/demois a sandbox: no key, no sign-up, the free tools only (design guide, compose a design, render, export PNG, share). Designs made there are public and may be removed; it is rate-limited per network. Claude Code:claude mcp add --transport http ridvay-demo https://mcp.ridvay.com/mcp/demo.Sign in (OAuth 2.1) — for clients that support MCP OAuth (claude.ai custom connectors, ChatGPT connectors/plugins, Claude Code, Cursor, MCP Inspector): add
https://mcp.ridvay.com/mcpwith no key, and the client discoversapi.ridvay.comas the authorization server, registers itself (Dynamic Client Registration) or presents a Client ID Metadata Document, and sends you toridvay.com/oauth/consentto approve. The token it receives is a normal Ridvay API key named " (OAuth)", so you can revoke it any time from your account page (or the client can call the revocation endpoint).
Living designs tools (0.6.0)
list_designs, share_design (view / edit / live-image / screen links), get_values and set_values
(update a design's declared {{variables}} — a menu board or price board that refreshes itself on the
/screen/<id> page and the /img/<id>.png live image), and list_brands (pass brand_id to generate_poster).
Related MCP server: KieAI MCP
How it works

Every frame of that GIF was designed and rendered by this MCP server — a four-page design
composed as design IR, saved with create_poster, then exported page by page with
export_poster. No screen recording, no image editor.
Tools
Tool | What it does |
| Start here — the design-IR authoring spec (element types, backgrounds, fonts, worked example) that teaches your assistant to compose designs itself. |
| The preferred path: your assistant composes the design as Ridvay design IR — Ridvay only stores, renders, and shares it. Fast, free, full creative control. |
| Fallback: Ridvay's AI designs from a text brief (slower, consumes the account's generation credits). |
| Turn an existing design image (poster photo, screenshot, export — file path or URL) into an editable design: text becomes editable, shapes become vectors, imagery is re-rendered fresh. Slow (1–4 min), uses generation credits. |
| Natural-language edit of an existing design ( |
| Report whether a design's AI images finished rendering and return its links + view count ( |
| Render a design to a downloadable PNG/JPEG at its native pixel size ( |
| Add motion (entrance/exit, page transitions, morph) — blank for a tasteful default, or describe it. |
| Render an animated design to a downloadable H.264 MP4 (optional looped soundtrack). |
| Poll an async |
Your AI client is the designer; Ridvay is the save/render/share backend. Client-authored
designs may still include prompt image slots — Ridvay renders those server-side after
creation. generate_poster remains for when the assistant can't compose the design itself.
Export & motion: export_poster gives you the actual poster image at its real dimensions
(e.g. a 1080×1350 PNG), not the share page. For animation, either include motion fields when
you compose the design (see the guide) or call animate_poster, then export_video for an MP4.
size accepts 1080x1080 (default), 1080x1920 / story, 1080x1350, 1920x1080,
a4, slide, or any WxH.
Live images
Every shared design also gets an always-current image URL — each fetch re-resolves the
design's live data bindings ({{time.now}}, countdowns, JSON feeds, declared {{vars}}),
so an embedded image never goes stale (re-rendered at most every 60 s):
https://ridvay.com/img/{designId}.pngDeclared vars are filled from query params, which makes personalized email images a one-URL job with your ESP's merge tags:
https://ridvay.com/img/abc123.png?name=*|FNAME|*A .jpg variant plus page, scale, w, and quality params are supported — see
get_design_guide ("Live data bindings + live image URL") for the bindings format (including
the reserved param names that can't double as binding/var keys).
Every live-image load and every /d/ share-page view counts toward the design's view
count — check_poster reports it as Views: N.
Get an API key
Sign in and create a key at ridvay.com/user/api-keys. Designs generated through MCP appear in your own My designs.
Quickstart
Claude Code
claude mcp add --scope user ridvay --env RIDVAY_API_KEY=sk-ridvay-… -- npx -y ridvay-mcpClaude Desktop — add to claude_desktop_config.json
(macOS: ~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"ridvay": {
"command": "npx",
"args": ["-y", "ridvay-mcp"],
"env": { "RIDVAY_API_KEY": "sk-ridvay-…" }
}
}
}VS Code / GitHub Copilot agent mode — add to your user mcp.json
(Command Palette → "MCP: Open User Configuration"), then enable ridvay in the
Copilot Chat tools picker:
{
"servers": {
"ridvay": {
"type": "stdio",
"command": "npx",
"args": ["-y", "ridvay-mcp"],
"env": { "RIDVAY_API_KEY": "sk-ridvay-…" }
}
}
}Then ask your assistant things like:
Generate a story-size poster for our weekend flash sale — 30% off everything, Saturday only.
Examples
A poster, straight from a brief
Generate a story-size poster for our weekend flash sale — 30% off everything, Saturday only.
A promo video, end to end — design, animation, and MP4 in one turn:
Using the ridvay MCP, create a 3-page 1080x1920 promo video design about [topic]. Page 1: a bold hook. Page 2: three or four key points with accent bars. Page 3: a worked example plus a call to action. Dark navy background, blue accents, Poppins headlines. Then animate it with snappy staggered entrances and morph transitions, and render it to MP4.
An existing design, made editable again
Recreate ~/Desktop/last-years-flyer.png as an editable design, then change the date to October 3rd and set the headline in our brand blue.
An image that never goes stale — live bindings resolve on every fetch:
Build a 1080x1080 status card with today's date and a countdown to our September 1 launch, then give me the live image URL.
A longer walkthrough of the video example, with the agent's actual tool sequence, is on the blog: Ridvay MCP: free AI design generation from Claude Code.
Environment
Var | Required | Meaning |
| yes | Your Ridvay API key ( |
| no | Default |
| no | Base for returned links, default |
| no | Admin/platform keys only: act on behalf of a specific user. |
Behavior notes
Sharing:
generate_poster/create_postercreate an unlisted public share link by default so the chat reply contains a working/d/{id}URL. Passshare: falseto keep a design private to your account (only the Studio edit link is returned).Deferred images: generation returns as soon as the layout is ready; AI/stock images render server-side in the background (~1 min).
check_posterreports on and, if needed, re-triggers that pass.
Telemetry
Content and usage data from the MCP (such as prompts and generated designs) may be used to
improve Ridvay's products, services, and AI features. Requests include basic attribution:
the connecting MCP client's name/version and, when the assistant provides the optional
agent_model argument, the model id that authored the design.
Development
npm install
npm run build # emits dist/
npm test # vitest unit suite
node dist/index.js # run the stdio server directly (needs RIDVAY_API_KEY)MIT © Ridvay
Available Tools
5 toolscheck_posterCheck a poster's statusA
Check whether a generated Ridvay design has finished rendering its AI images, and get its view/edit links again. Use after generate_poster reports images still rendering.
Inputs: design_id (required).
create_posterCreate a poster from your own designA
Save a poster design that YOU (the assistant) composed as Ridvay design IR — you control every element, color, and font; Ridvay only stores, renders, and shares it. No Ridvay-side AI generation is involved. Call get_design_guide first for the IR format. Use generate_poster instead when Ridvay's AI should do the designing. Returns view/share/edit links.
Inputs: design (required), share.
generate_posterGenerate a poster with Ridvay StudioA
Generate a poster, flyer, social-media post, story, banner, or any graphic design from a text brief using Ridvay Studio's AI design engine. Returns links to view, share, and edit the design. Describe the content, occasion, style, and any text that must appear. Typical run time is 20–40 seconds.
Inputs: prompt (required), size, share, use_brand.
get_design_guideGet the Ridvay design-authoring guideA
Returns the Ridvay design IR format (JSON contract, element types, fonts, backgrounds, worked example) so YOU can compose a poster design yourself and save it with create_poster. Always call this before your first create_poster call.
Inputs: no named parameters.
refine_posterRefine an existing posterA
Edit a previously generated Ridvay design with a natural-language instruction (e.g. "make the headline red", "add our opening hours", "swap to a dark theme"). Requires the design ID returned by generate_poster.
Inputs: design_id (required), instruction (required), use_brand.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Each tool has a clearly distinct purpose: get_design_guide provides the format for manual design, create_poster saves a manual design, generate_poster creates with AI, check_poster checks rendering status, and refine_poster edits existing designs. No overlap in functionality.
All tool names follow consistent verb_noun snake_case pattern (e.g., check_poster, create_poster, generate_poster, get_design_guide, refine_poster). No mixing of conventions or styles.
With 5 tools, the server is well-scoped for its purpose. Each tool is necessary for the core workflow of creating and managing graphic designs, without bloat or deficiency.
The tool surface covers the main lifecycle: creation (manual and AI), status checking, and editing. Missing a delete or list tool is a minor gap, but the core design workflow is complete.
Related MCP Connectors
MCP server for Midjourney AI image generation and editing
An MCP server that integrates with Discord to provide AI-powered features.
The Canva MCP server connects AI assistants (like Claude, ChatGPT, and Cursor) to Canva's API, enabling them to create and manage designs directly within chat conversations. Key capabilities include generating new designs from prompts, autofilling templates, searching and resizing existing designs, importing files from URLs, exporting designs as PDFs or images, and managing folders and comments without switching between tools.
MCP server for Flux AI image generation
Related MCP Servers
AlicenseAqualityFmaintenanceAn MCP server that integrates with Recraft AI to enable generation and manipulation of high-quality raster and vector images through tools like image generation, editing, vectorization, background removal, and upscaling.9305 npm60MIT- AlicenseAqualityDmaintenanceMCP server for AI-powered media generation: images, videos, audio, and upscaling using 99 AI models.6MIT
- AlicenseAqualityDmaintenanceMCP server for AI image generation and editing using Google Gemini image models.71,647 npmMIT
- AlicenseAqualityBmaintenanceGenerate AI images, video, voiceovers and music from Claude, Cursor or VS Code through 50+ models (Veo 3.1, Kling 3, Seedance, Nano Banana, GPT Image, ElevenLabs). Upload photos for edits and image-to-video. Tools are annotated (read-only vs. credit-spending); failed generations are refunded. Local via npx or hosted with OAuth at fattly.app.651 npmMIT