1. Conduid
  2. IoT
  3. Scry MCP
MCP server · IoT

Scry MCP

Computational scrying glass — MCP server for visual computing with embedded Python, resvg rendering, and live web gallery

Unclaimed iot
34Low

Scored 6 months ago · breakdown

About Scry MCP

Scry MCP is an MCP server in the IoT category: computational scrying glass — MCP server for visual computing with embedded Python, resvg rendering, and live web gallery.

Install

From source
git clone https://github.com/tobert/scry-mcp

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about Scry MCP

Powered by Claude · Grounded in docs

I know everything about Scry MCP. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

  • ✓README presentThe repository ships documentation.

5 checks have not been run yet.

README

Computational scrying glass — an MCP server that lets Claude generate SVG visuals via Python code, render them to PNG, and serve them in a live web gallery.

Architecture

Tools

whiteboard — Execute Python code to generate SVG visuals on a named board. Call svg('<svg>...</svg>') to set content. Variables persist between calls.

whiteboard_list — List all active boards with thumbnails, URLs, and metadata.

Requirements

  • Rust 1.85+ (edition 2024)
  • Python 3.x with development headers (PyO3 embeds CPython at build time)
  • System fonts — resvg uses fontdb to load system fonts for text rendering

Built and tested on Arch Linux (kernel 6.18, Python 3.14, Rust 1.90). Should work on other Linux distros and macOS but hasn't been tested yet — reports welcome.

Install

cargo install scry-mcp

Or build from source:

git clone https://github.com/tobert/scry-mcp
cd scry-mcp
cargo build --release

On Debian/Ubuntu you may need python3-dev for PyO3:

sudo apt install python3-dev

Add to Claude Code

claude mcp add -s user scry -- /path/to/scry-mcp --address 127.0.0.1 --port 3333

Or if installed via cargo install:

claude mcp add -s user scry -- scry-mcp --address 127.0.0.1 --port 3333

Or manually add to your MCP config:

{
  "mcpServers": {
    "scry": {
      "command": "scry-mcp",
      "args": ["--address", "127.0.0.1", "--port", "3333"]
    }
  }
}

Usage

Once configured, ask Claude to use the whiteboard tool:

"Draw a red circle on a white background using the whiteboard tool"

The gallery is live at http://localhost:3333/gallery/ — it auto-refreshes via SSE when boards update.

CLI Options

scry-mcp [OPTIONS]

Options:
      --address <ADDRESS>  Gallery bind address [default: 127.0.0.1]
      --port <PORT>        Gallery port [default: 3333]

Python Environment

Each board gets a persistent Python namespace with these pre-imported:

math, random, json, re, textwrap, itertools, functools, collections, colorsys, hashlib, string, dataclasses

WIDTH and HEIGHT are set to board dimensions (default 800x600).

Dangerous modules (os, subprocess, socket, etc.) are blocked.

Examples

All of these were generated by Claude using scry's whiteboard tool.

Data Visualization

Generative Art

Simulation

README copy captured when this server was indexed; it may lag the repository. The original is authoritative.

Questions

About Scry MCP

How do I install Scry MCP?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Scry MCP safe to use with an AI agent?

Its trust score is 34 out of 100 (low). It passes 1 of the 1 static security checks run so far. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Scry MCP still maintained?

Conduid hasn't recorded a commit date for this repository yet. Check the repository directly for recent activity.