mcpgraph

Build no-code MCP servers using YAML graphs with JSONata and JSON Logic.

18|2|Updated Dec 18, 2025
One-click install
npx skills add https://github.com/TeamSparkAI/mcpGraph --skill mcpgraph
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mcpgraph
Source: https://github.com/TeamSparkAI/mcpGraph/tree/main/skills/mcpgraph
Command: npx skills add https://github.com/TeamSparkAI/mcpGraph --skill mcpgraph

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Build no-code MCP servers with tools that compose and orchestrate other MCP tools.

Core Features & Use Cases

  • Declarative YAML-based server graphs that wire tools across MCP servers.
  • Data transformation with JSONata and routing decisions with JSON Logic for observable workflows.
  • Use Case: Orchestrate a sequence of tool calls, transform intermediate results, and branch logic without writing code.

Quick Start

Install mcpgraph locally and define a YAML graph as shown in the examples; then run the mcpgraph command with your graph to start the server.

Frequently Asked Questions about mcpgraph

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I orchestrate multiple MCP tools without writing custom code?

MCP server orchestration without code is achieved by defining declarative YAML graphs with mcpgraph. You model tool calls, transform intermediate data with JSONata, and route decisions using JSON Logic across connected MCP servers.

What is a declarative YAML graph for MCP server composition?

A declarative YAML graph for MCP server composition is a configuration file defining server, mcpServers, and tools. It uses entry, mcp, transform, switch, and exit nodes to wire tool calls and route decisions across servers without embedding code.

How do I transform data between MCP tool calls using JSONata?

You transform data between MCP tool calls by defining transform nodes in your mcpgraph YAML definition. These nodes apply JSONata expressions to intermediate results, allowing you to reshape data payloads before passing them to subsequent tool calls.

Can I route decisions between MCP tools without embedding logic in my application code?

Yes, you can route decisions without embedding logic by using switch nodes in your mcpgraph YAML definition. Switch nodes evaluate conditions using JSON Logic to branch workflow paths dynamically based on intermediate tool outputs.

Do I need to install any specific software to build no-code MCP servers?

Yes, you need to install the mcpgraph tool locally on your machine. After installation, you define your YAML graph with server definitions and run the mcpgraph command with your graph file to start the server.

What are the limitations of using declarative graphs for MCP data orchestration?

Declarative MCP data orchestration relies entirely on JSONata for transformation and JSON Logic for routing. Complex workflows requiring custom imperative logic or external library dependencies may exceed the constraints of these configured entry, mcp, transform, switch, and exit nodes.