MCP Contract

A specification for composable AI workflow bundles with typed contracts between layers.

v0.1.0-draft Apache-2.0 JSON Schema draft 2020-12

MCP Prompts    = source code (operational reasoning, portable)
MCP Tools      = runtime libraries (state + execution, portable)
MCP Apps       = UI framework (interactive views, portable)
Platform skill = compiler flags (thin per-platform wrapper)
Claude/ChatGPT = the compiler

Prompts ──provides──> ["analysis-request"]
                            |
Tools   ──provides──> ["sec-filings", "comps-table"]
                            |
Apps    ──consumes──> ["comps-table", "analysis-request"]

Skills  ──targets───> ["claude", "gpt"]

Compiler (Claude, GPT, Gemini, Llama) links it all at runtime.

Prompts

Reasoning logic. What to think about and what output shape to produce.

analogous to source code

Tools

Runtime execution. APIs, databases, computation engines with typed schemas.

analogous to runtime libraries

Apps

Interactive views. Dashboards, reports, forms that consume contract schemas.

analogous to UI frameworks

Skills

Platform adaptation. Thin wrappers for Claude, GPT, Cursor, etc.

analogous to compiler flags

Get Started

# Install the CLI pip install mcpc-cli # Scaffold a new bundle mcpc init my-workflow # Validate contracts mcpc validate # Pack for distribution mcpc pack

Why?

MCP servers are monoliths. To use one server's prompts with another's tools and a third's UI, you fork all three. MCP Contract separates bundles into independently authored, versioned, and composable layers wired together through typed contracts.

No custom schema language. No new runtime. Just a manifest format that makes composition, package managers, and marketplaces possible.

Reference Implementation

Synth Ops is an intelligent infrastructure operations system built as an MCP Contract bundle.

Schema

The manifest schema is hosted at mcpcontracts.com/schema/0.1.0.json for editor validation.