SeanPropApp MCP Server

SeanPropApp Methodology v2.1.0 / MCP protocol 2025-06-18

A 19-module proposition analysis methodology, available two ways: as a guided browser app, and as a Model Context Protocol server that any MCP-capable AI client can call directly inside the user's host. We recommend the browser app for the full guided flow that produces polished documents to share with others, and MCP for invoking individual modules inside your own AI or agentic session.

Ready to connect? Step-by-step setup for every host is at /mcp-setup. Prefer the guided browser experience? Start at prop.seanoneill.com.

What it does

The SeanPropApp MCP server publishes Sean O'Neill's 19-module proposition analysis methodology as MCP tools. An AI host such as Claude, ChatGPT, Cursor, or Claude Code calls a tool, our server assembles the system prompt and user message for that module on the methodology version of record, and returns the prompt block. The host then runs inference locally with its own model and API key. We never call an LLM, never store the host's output, and never see the conversation that follows.

The same methodology runs the browser app at prop.seanoneill.com, so output structure is identical across surfaces.

Available tools

run_module

Run a single methodology module on a company. Returns a fenced system prompt and user message block that the host then executes with its own model. Accepts a module ID, the company / initiative context, and an optional map of prior module outputs for downstream modules that depend on earlier results.

list_modules

List all 19 methodology modules with their IDs, display names, phase (foundation / positioning / extended), and word limits. Call this first to discover valid module_id values for run_module.

Methodology

Read the full module list, phasing, and design rationale on the methodology page. The MCP server pins methodology version v2.1.0; every run_module response includes this version in its provenance footer so downstream artifacts are reproducible.

Demo analyses

Four public analyses generated by running the full methodology in Auto-Run mode. Same prompt assembly the MCP server uses, on public information only.

New Product Idea
DocuSign AI Contract Negotiation Workspace
Could DocuSign move beyond signature and become more central to contract decision-making?
Hypothetical M&A
Pendo Acquires LaunchDarkly
Would this deal create a stronger product platform and a more defensible strategic position?
New Market
Tesla Optimus: Elder Care Robotics
Can Tesla translate manufacturing scale and AI into a trusted, safety-certified eldercare robot?
Whole Company Analysis
Cursor
Can an AI-native coding product build durable advantage as code itself gets cheaper?

How to connect

Step-by-step setup for both the remote MCP path (custom connector in Claude, ChatGPT, Cursor) and the local stdio path (npm package + personal access token) lives at /mcp-setup.

Privacy and data handling

Our server assembles prompts only; it never calls a language model, never stores your host's output, and never sees the conversation that follows a tool call. Telemetry is content-free metadata only. Full data-handling statement at /privacy.