MCP Server · Free to try

Turn any AI agent into
a visual designer.

Edithly's MCP server gives Claude, Cursor, and any MCP-compatible agent 21 visual formats to build with — infographics, presentations, mind maps, flashcards, and more — generated from natural language, without leaving your agent.

Claude
Cursor
Windsurf
Zed

No credit card required  ·  20 free credits  ·  Setup in under 2 minutes

Product demo

Claude generating an infographic via Edithly MCP

Schema-driven, every time

Structured output instead of freeform hallucinated HTML.

The agent picks the format

Claude or Cursor reasons about your content on its own.

Never leave your agent

Ask, review, and iterate without a separate tab.

Ask in plain language

Turn this research doc into an infographic with the top 3 stats and a comparison table.

Your agent picks the format, fills the schema, and hands back a link — no design tool, no copy-pasting.

21 formats

One tool call away from any format

Every format is schema-driven — the agent fetches the exact structure before it builds, so output stays consistent instead of improvised.

Teach & study

  • Flashcard deck
  • MCQ quiz
  • Question paper
  • Study aids
  • Solvesheet
  • Glossary
  • Cheatcode

Explain & visualize

  • Infographic
  • Mind map
  • Timeline
  • Comparison table

Present & pitch

  • Slide presentation
  • One-pager
  • Brochure
  • Case study
  • Showcase

Plan & execute

  • Action plan
  • Use cases
  • ROI calculator
  • Kanban board
  • AI poster image
ClaudeCursorWindsurfZed

How the agent decides

Three tool calls, no guesswork

This is the exact sequence Edithly's MCP server tells the agent to follow, in whichever client you're already working in.

01

The agent lists what's possible

Your agent calls get_visual_types() and sees all 21 formats — infographic, presentation, mind map, flashcard deck, and more — each with a description of what it's best for.

02

It fetches the exact schema

get_visual_schema(visual_type) returns the precise structure and a worked example, so the agent fills in real content instead of guessing at layout.

03

It renders and hands back a link

create_visual() (or create_image / create_solvesheet for those types) renders the design server-side and returns a shareable Edithly link — no copy-pasting HTML back into chat.

See it in action

A real request, start to finish — inside Claude, no tab switching.

Full walkthrough

Ask → agent picks a format → visual link comes back

Connect in under 2 minutes

Add this to your Claude Desktop, Cursor, or other MCP client config file. Swap in the API key from your Edithly account settings.

{
  "mcpServers": {
    "edithly": {
      "command": "npx",
      "args": ["-y", "@edithly/mcp-server"],
      "env": { "EDITHLY_API_KEY": "your-api-key" }
    }
  }
}

Restart your AI tool after saving — Edithly's tools will appear in its available tool list.

What people build

What people build with it

01

Turn research into something visual

Drop a messy doc or web page into the conversation and ask for an infographic, mind map, or comparison table. The agent reads the content and calls create_visual() with the right schema — no design tool required.

Turn research into something visual

Screenshot or short clip of the result

02

Generate study material on demand

Ask for a flashcard deck, MCQ quiz, or question paper from whatever your agent already has open. Solvesheets are backend-solved, not agent-invented, so worked steps stay accurate.

Generate study material on demand

Screenshot or short clip of the result

03

Ship a deliverable, not just an answer

Have the agent draft a one-pager, brochure, or slide presentation it can hand straight to a teammate or prospect — a link back, not a wall of markdown.

Ship a deliverable, not just an answer

Screenshot or short clip of the result

Why this beats another chat window

21 formats, one schema-driven tool

Every visual type — jumbotron, stats grid, chart, table, timeline, kanban — follows a defined component schema, so output is structured and on-brand instead of freeform hallucinated HTML.

The agent picks the right format

Claude or Cursor reasons about your content and chooses cheatcode vs. infographic vs. one-pager on its own, then fills in the schema — you don't have to know Edithly's format names.

Never leave your agent

No browser tab, no separate app to open. Ask for a visual, review it, ask for changes — all inside Claude Desktop, Claude Code, Cursor, or Windsurf.

Projects and credits, tracked

list_projects() and get_credit_balance() let the agent check what's already been made and how much budget is left before it generates something new.

Give your agent a design studio

Create a free Edithly account, grab your API key, and your agent can start generating visuals in minutes.

Frequently asked questions

What is MCP?+

The Model Context Protocol (MCP) is an open standard, introduced by Anthropic, that lets AI agents like Claude connect to external tools through one common interface instead of a custom integration for every tool.

What can I actually generate through Edithly's MCP server?+

21 visual types: infographics, mind maps, slide presentations, flashcard decks, MCQ quizzes, question papers, comparison tables, timelines, brochures, one-pagers, case studies, action plans, ROI calculators, kanban boards, AI poster images, and more. Call get_visual_types() from your agent for the full list with descriptions.

Which AI tools support MCP?+

Claude Desktop, Claude Code, Cursor, Windsurf, Zed, and a growing list of MCP-compatible clients. Any agent built on the MCP standard can connect to Edithly's MCP server.

Do I need to be a developer to set this up?+

No. Setup is a short JSON config entry (shown above) in your AI tool's settings. You'll need a free Edithly API key from your account — no coding required for basic use.

How much does it cost?+

Free accounts start with 20 credits. Most visual types cost 1 credit each; AI poster images cost 4 credits. Unused credits roll over — see the app for current plan details.

Does the agent make up the content, or is it grounded?+

The agent supplies the content based on what's in its context — your documents, chat history, or research. Solvesheets are the one exception: the backend solves the problem server-side so the agent can't invent steps.