Axiocairn MCP

Axiocairn MCP

Checks portfolio risk for your holdings from an AI chat: concentration, volatility and beta to SPY, plus scenarios under stated assumptions. A hosted MCP server for research, not investment advice.

Made by cplj on GitHub. These notes are Mindket's, written from its README; the project's own files and license are in its repository.

About

What it is: Axiocairn MCP links an AI chat app to a hosted service that checks the risk in your holdings. It measures how concentrated they are, their historical volatility and their beta to SPY, which estimates how sensitive they are to SPY. It also runs scenarios under stated assumptions and reads dated research that your account can see.

Who it suits: It suits people who hold a handful of positions and want to see where their risk sits before they think about changes. Analysts and students can also use it to practise reading beta and concentration.

Industry | Who uses it | How they use it | Example
Personal investing | Investors with a few holdings | See where portfolio risk piles up before weighing any change | A portfolio of NVDA, MSFT and SPY
Investment research | Analysts | Read the dated research that their account can see | Asking which reports are available today
Finance education | Students and teachers | Practise reading beta and concentration on made-up holdings | A made-up holding with fixed weights

You can skip it if you want trade signals or a broker link, since it offers no trade execution and no advice.

Where to use it:

• Claude on the web: the project checked its connector on October 27, 2026
• ChatGPT on the web: checked on the same date, with its own connector address
• Grok and Muse: checks are still pending
• Claude Code, Cursor and Gemini CLI: the README gives configuration templates, and checks are pending

Use case | Who | What they get back
Concentration check | Investors | The largest position, the technology weight and a concentration score
Market sensitivity | Investors and analysts | Daily volatility, beta to SPY and beta on days when SPY fell
Report lookup | Analysts | A list of today's reports and how recent the newest one is

How to set it up: You need a few things in place first.

• An AI app that lets you add a custom connector, such as claude.ai
• A Google or GitHub account to sign in with
• An account plan, because access follows your plan

1. Open your AI app's connector settings and create a custom connector called Axiocairn. Use the server address listed in the project's README.
2. For ChatGPT, use the separate ChatGPT address that the README lists.
3. Log in through your Google or GitHub account, and read the access approval screen before you accept.
4. Ask which Axiocairn reports are available today to confirm it works.

No local install is needed, and no API key goes into a Cursor configuration. The Claude Code, Cursor and Gemini CLI templates in the README are not verified by the project.

How to use it: Ask in plain language about your holdings or about the reports. These three questions from the README are a good first try.

What Axiocairn reports are available today?
What is the data date of the latest Axiocairn report you can see, and is it today's edition?
In a few short bullets, what are the limits of Axiocairn reports?

Developers can call the server directly. A minimal request that asks for report timing looks like this.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_report_schedule",
    "arguments": {"language": "en"}
  }
}

Tool | What it does
tools/list | Lists the tools available now, each with a description and its input format
get_report_schedule | Returns report timing when it is called through tools/call

Example: Ask how exposed a portfolio of half NVDA, 30% MSFT and 20% SPY is to the market. The README shows the answer as a set of readings from data dated 2026-10-30, over 252 daily returns.

Reading | Value
Largest position | NVDA at 50%
Technology weight from NVDA and MSFT | 80%
Concentration (HHI) | 0.38
Daily volatility | 1.58%
Beta to SPY | 1.47
Beta on days when SPY fell | 1.58, over 121 of 252 days

The holding is made up, with fixed weights. It is no real account, and nothing in it is advice. The readings show how the three symbols moved in that window. They are not a forecast.

For newcomers and experts: If you are new, add the connector, sign in and ask the three questions above. They show which reports exist, how fresh the latest one is and where its limits lie. Nothing here needs a broker.

If you know the field, note how it connects. The server is hosted and speaks Streamable HTTP with an account authorization flow, so nothing runs locally. After you authorize, call tools/list to see the current tool names and schemas. When you share an answer, include the report link, the data date, any fields that were missing and the stated limits.

Add the custom connector
Log in with Google or GitHub
Ask about holdings or reports
Read the data date and limits

Good to know:

• Treat results as general research only. They are not tailored to you, and they do not tell you to buy or sell
• There is no brokerage connection and no trade execution
• A result changes with the time window, the data coverage and the assumptions, and past relationships may not hold
• A scenario answers what-if questions on assumptions you state, and it predicts nothing
• Your account allowance pays for some calculations and for making new reports, so check the inputs and the allowance first
• To save, edit or remove alerts, research plans or scenarios, you need the matching access and must accept the user agreement
• Use synthetic or non-identifying inputs rather than real account details

On GitHub, 7 November 2026

• Stars: 86
• Forks: 0
• Open issues and pull requests: 0
• Language: Python
• License: MIT
• Topics: ai-agents, ai-tools, backtesting, chatgpt, claude, finance, investing, llm, mcp, mcp-server, model-context-protocol, quant

Links

• Code: https://github.com/cplj/axiocairn-mcp
• Website: https://axiocairn.com/
• Releases: https://github.com/cplj/axiocairn-mcp/releases
• Issues: https://github.com/cplj/axiocairn-mcp/issues
• Star history: https://www.star-history.com/#cplj/axiocairn-mcp&Date

Get the code

git clone https://github.com/cplj/axiocairn-mcp.git

Then follow the setup in its README. Nothing in this kit runs by itself.