Skip to content
JumbleFixJumbleFix
MCP

Install JumbleFix MCP

Last updated: August 14, 2026

Colorful workspace featuring a laptop with code editor and vibrant screens.
Photo by Daniil Komov on Pexels

Add JumbleFix to your coding agent so it can unscramble letter racks and look up dictionary senses. No API key. Works with Cursor, VS Code Copilot, Claude Code, Windsurf, and Zed over the public /mcp endpoint.

endpoint

HTTP
https://jumblefix.com/mcp

Discovery: /.well-known/mcp/server-card.json

Choose your editor

Pick Cursor, VS Code, Claude Code, Windsurf, or Zed. Each tab has the matching install command or config.

Install for Cursor

One-click install or paste into mcp.json.

Install in Cursor
  1. Option A: click Install in Cursor to open the install prompt.
  2. Option B: press ⌘ShiftP (Mac) or CtrlShiftP (Windows/Linux), run MCP: Open MCP Settings, and add the JSON below to mcp.json.
  3. For local development, change the URL to http://localhost:5173/mcp (or your app port).

mcp.json

JSON
{  "mcpServers": {    "jumblefix": {      "url": "https://jumblefix.com/mcp"    }  }}
Cursor MCP docs

Showing setup for Cursor.

Ask the agent

After the server connects, send a prompt that names the tools:

prompt-1.txt

Text
Unscramble the letters listen with the JumbleFix MCP tools.

prompt-2.txt

Text
Use lookup_word to define silent.

prompt-3.txt

Text
Find 5-letter words from the rack aeiou using unscramble.

Tools

Exposed over MCP as unscramble and lookup_word. Blank tiles: ? or *.

tools/list result

JSON
{  "tools": [    {      "name": "unscramble",      "description": "Find valid words that can be made from the given letters.",      "inputSchema": {        "type": "object",        "required": ["letters"],        "properties": {          "letters": { "type": "string" },          "minLength": { "type": "integer" },          "maxLength": { "type": "integer" }        }      }    },    {      "name": "lookup_word",      "description": "Look up definitions for a word.",      "inputSchema": {        "type": "object",        "required": ["word"],        "properties": {          "word": { "type": "string" }        }      }    }  ]}

tools/call unscramble

JSON
{  "jsonrpc": "2.0",  "id": 3,  "method": "tools/call",  "params": {    "name": "unscramble",    "arguments": {      "letters": "listen",      "minLength": 3,      "maxLength": 6    }  }}

tools/call lookup_word

JSON
{  "jsonrpc": "2.0",  "id": 4,  "method": "tools/call",  "params": {    "name": "lookup_word",    "arguments": {      "word": "silent"    }  }}

curl (JSON-RPC)

Prefer MCP from an agent. Use these shells when debugging POST /mcp:

initialize.sh

Bash
curl -s https://jumblefix.com/mcp \  -H 'Content-Type: application/json' \  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}'

tools-list.sh

Bash
curl -s https://jumblefix.com/mcp \  -H 'Content-Type: application/json' \  -d '{"jsonrpc":"2.0","id":2,"method":"tools/list"}'

unscramble.sh

Bash
curl -s https://jumblefix.com/mcp \  -H 'Content-Type: application/json' \  -d '{    "jsonrpc":"2.0","id":3,"method":"tools/call",    "params":{      "name":"unscramble",      "arguments":{"letters":"listen","minLength":3,"maxLength":6}    }  }'

lookup.sh

Bash
curl -s https://jumblefix.com/mcp \  -H 'Content-Type: application/json' \  -d '{    "jsonrpc":"2.0","id":4,"method":"tools/call",    "params":{      "name":"lookup_word",      "arguments":{"word":"silent"}    }  }'

MCP vs HTTP API vs WebMCP

Use MCP when your agent already speaks MCP.

Use POST /api/unscramble and GET /api/words/{word} (see /openapi.json) for plain REST and richer filters.

Use WebMCP in browsers that support navigator.modelContext; jumblefix.com registers unscramble_letters and lookup_word automatically.

Auth is public/anonymous. See /auth.md if an agent needs discovery metadata.