just-signal-mcp

Just Signal MCP Server

Generate, ground, and publish Just Signal macro-intelligence briefings from inside the AI you already pay for — no API key entered into Just Signal.

Your model (Claude, ChatGPT, Copilot, Grok, Gemini, or any MCP-capable client) does the reasoning; Just Signal supplies what it uniquely owns: live Day-0 price grounding, mechanical dollar-target math, server-side report rendering (with the Scorecard data island + transmission stamps), publishing to the public library, the registry-driven Signal Scorecard, and The Chophouse panel grade. Because your subscription does the thinking, there is no per-call API key and no token cost inside Just Signal.

Tools

Tool	What it does
get_stock_quote	Live price + % change via Just Signal's Finnhub relay — establishes a ticker's Day-0 baseline.
publish_report	The server grounds every ticker's Day-0, computes dollar targets from your % projections, renders the report, and publishes it. Open to the community: tokenless, rate-limited per day; an owner token lifts the caps. Returns the URL.
list_reports	Recent reports in the public library.
get_scorecard	Grades every registry ticker with the unified verdict cascade (projection › direction › return) against the daily snapshot close — matches the website exactly, zero live quotes. Folds in the Chophouse doneness grade where a ticker is plated.
get_panel_grade	The Investment Panel's read on one ticker: its Chophouse doneness grade (RAW…CHARRED) plus every seat's vote across 11 emulated investor lenses. The panel advises; the reader decides.

Two ways to connect

A) Hosted URL — no install (most AI apps)

Add this as a custom MCP connector wherever your app configures them:

https://www.ou-reports.com/api/mcp.php

All five tools, including open community publishing (rate-limited, no key). Full per-vendor guide: https://www.ou-reports.com/mcp.html

B) Run it locally (Claude Desktop / Code, Cursor, LM Studio, …)

Requires Node 18+.

Once published to npm, the one-liner (no download):

claude mcp add just-signal -- npx -y just-signal-mcp

Or from a copy of this folder:

cd just-signal-mcp && npm install

Claude Desktop claude_desktop_config.json (macOS ~/Library/Application Support/Claude/, Windows %APPDATA%\Claude\):

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

(or "command": "node", "args": ["/ABSOLUTE/PATH/TO/just-signal-mcp/index.js"]). Restart the client; the five tools appear.

Optional environment variables

Var	Default	Purpose
JS_BASE_URL	https://www.ou-reports.com	Point at a different Just Signal host.
JS_PUBLISH_TOKEN	(none)	Owner token; bypasses the community publish rate limits. Not required to publish.

How to drive it

"Research grid-scale energy storage, then write me a Just Signal briefing: 5 trends, each with a hook, qualitative + quantitative analysis, a lead ticker in assetOrSector, and 1/5/10-year targets as percentages. Use get_stock_quote to sanity-check each Day-0 price, then publish_report."

Your model does the research and writing; the server grounds the prices, computes the dollar targets, stamps and renders the report, and publishes it. Ask for get_scorecard to see how past calls are holding up, or get_panel_grade NVDA for the panel's steakhouse read on a name.

Targets must be percentages (+18%) or multiples (2.5x), never dollars — the server computes the dollar figure from the live Day-0 price. Private / no ticker → use (PRIVATE) and the trend honestly shows relative basis.

Publishing is open — and safe

Publishing needs no key and no token: the server itself grounds every price and renders every page, so the only thing anyone can add to the library is a genuine Just Signal report — never arbitrary HTML. To keep it fair, tokenless publishing is rate-limited (a few reports per connection per day, plus a global daily budget); read tools are never limited. This server never sees or transmits any AI API key; it only talks to Just Signal's own endpoints over HTTPS.

Not investment advice

Just Signal is a futurism instrument. Projections are model hypotheses from a stated Day 0; the Scorecard and the Chophouse exist to measure them honestly — including the misses.