prettify-mcp

An MCP server that turns a pasted payload into a shareable Prettify link. Ask your AI coding tool to "visualize this JSON" and get back a prettify.tools/s/… URL that opens the data in an interactive viewer (tree, table, chart, or diagram).

Hosted docs: https://prettify.tools/docs/api

Tools

| Tool | Purpose | Returns | |---|---|---| | prettify_view({ content, format? }) | Open a payload in Prettify | share URL | | prettify_query({ content, filter, format? }) | Run a jq filter over the data | result (JSON) + view link | | prettify_convert({ content, from, to }) | Convert between json/yaml/csv/xml | converted text + view link | | prettify_diff({ left, right, format? }) | Structural diff as an RFC 6902 JSON Patch | patch + summary + view link | | prettify_infer_types({ content, target, format? }) | Generate types from a sample | code (+ view link for json-schema) |

format is one of json (default), yaml, csv, xml, mermaid (mermaid is view-only; convert/diff accept the four data formats). prettify_infer_types target is one of typescript, go, kotlin, rust, json-schema.

The transform tools reuse Prettify's own engines (the jq implementation, format converters, semantic diff, and type inference), and each returns the data result plus a Prettify link to view it. Anonymous by default (no account, no key): the view links call Prettify's public REST endpoint, which applies a small free-tier size cap and per-IP rate limit; a Pro license key raises those limits.

Requirements

• Node.js 18+

Configuration

| Env var | Default | Purpose | |---|---|---| | PRETTIFY_API_URL | https://prettify-api.alexi-81a.workers.dev | API base URL (override to point at a local worker or proxy) | | PRETTIFY_LICENSE_KEY | (none) | Optional Pro license key; forwarded as a bearer token to raise size/rate limits |

Client setup

Claude Desktop

Add to claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "prettify": {
      "command": "npx",
      "args": ["-y", "prettify-mcp"]
    }
  }
}

With a Pro license key:

{
  "mcpServers": {
    "prettify": {
      "command": "npx",
      "args": ["-y", "prettify-mcp"],
      "env": { "PRETTIFY_LICENSE_KEY": "XXXX-XXXX-XXXX-XXXX" }
    }
  }
}

Restart Claude Desktop. The prettify_view tool then appears in the tools list.

Claude Code

claude mcp add prettify -- npx -y prettify-mcp

Or add to .mcp.json in your project:

{
  "mcpServers": {
    "prettify": { "command": "npx", "args": ["-y", "prettify-mcp"] }
  }
}

Cursor

Add to ~/.cursor/mcp.json (global) or .cursor/mcp.json (per-project):

{
  "mcpServers": {
    "prettify": { "command": "npx", "args": ["-y", "prettify-mcp"] }
  }
}

Example

You: Visualize this as a chart: [{"month":"Jan","sales":120},{"month":"Feb","sales":180}]

Assistant: (calls prettify_view) → https://prettify.tools/s/1a2b3c4d5e6f

Development

npm install
npm test              # vitest, mocked fetch
npm run build         # tsc -> dist/

# Interactive testing with the MCP Inspector (https://modelcontextprotocol.io/docs/tools/inspector):
npm run inspect       # builds, then opens the Inspector UI (uses the default/prod API)
npm run inspect:local # same, but PRETTIFY_API_URL=http://localhost:8788 (run the worker first)

Add -e PRETTIFY_LICENSE_KEY=... to the Inspector command to test the Pro tier.

Privacy

The content you pass is compressed and stored behind a short link so the viewer can load it. Only origin, format, and byte-size are recorded for analytics; never the content itself.