AI Tools Hub

Claude Stock-Analysis Prompts: What's Actually Official

A post claiming Anthropic "officially released 6 stock prompts" has been circulating on social media. There's no such document. Here's what Anthropic has actually published for financial analysis, and six prompt templates you can build yourself using the same structure.

Disclosure: There are no affiliate or sponsored links on this page. Links to Claude, Anthropic, and the public data portals mentioned below are plain informational links, not affiliate links.

Quick answer

There is no official Anthropic list of "6 Claude stock prompts." What Anthropic has actually shipped is a prompting-strategies guide for financial analysis and an open-source Financial Services plugin set (comparable-company analysis, DCF models, LBO models, and earnings-transcript analysis, run with slash commands like /comps, /dcf and /earnings) built for Claude Cowork and Claude Code. It targets finance professionals with paid institutional data connections, not a walk-up feature in the consumer app. The useful part for an individual investor is the underlying prompt structure — company, metric, period — which you can reuse by hand.

What's official vs. what you build yourself

Two real Anthropic resources exist. Neither is a plug-and-play consumer stock screener.

Prompting strategies for financial analysisAnthropic's own tutorial on structuring finance prompts for Claude — general guidance, not a numbered prompt list.
View tutorial →
Financial Services plugins (open source)github.com/anthropics/financial-services — comps, DCF, LBO and 3-statement workflows for Claude Cowork/Code, with connectors to institutional data providers.
View repository →

The plugin set's core "Financial Analysis" plugin is the foundation the others build on, and it's explicitly aimed at workflows like comparable-company analysis, DCF and LBO modeling and 3-statement financials, wired up to data providers such as Daloopa, Morningstar, S&P Global, FactSet, Moody's and PitchBook. That's an enterprise research-desk setup, not something that turns on inside claude.ai for a retail account. If you're analyzing a stock in the ordinary Claude app or website, none of those connectors are available to you — you're supplying every number yourself.

What still carries over is the shape of a good finance prompt. Below are six prompt angles that map to the kinds of questions individual investors actually ask, each rewritten around a simple structure: company + metric + period. State those three things explicitly and Claude has far less room to guess wrong.

Six prompt templates, company + metric + period

1. Earnings check

Closest official equivalent: the equity-research earnings-transcript workflow (/earnings)

Use this after a quarterly report or call. Paste the release or transcript excerpt — Claude cannot see it otherwise.

For [Company Name, ticker], using the attached [Q_ fiscal year] earnings release: summarize [revenue and operating margin] versus the same quarter last year, and list every number as (a) reported figure, (b) year-over-year change, (c) one line on what management said caused the change. Flag anything you can't find in the text instead of estimating it.

2. Competitor comparison

Closest official equivalent: comparable-company analysis (/comps)

Name every company being compared — Claude should never be left to decide who counts as a "competitor."

Compare [Company A] against [Company B and Company C] on [gross margin] for [fiscal year 2025], using the figures I've pasted below for each. Output a table with one row per company and one column per period. Do not fill in a number you weren't given.

3. Sector/industry comparison

Closest official equivalent: comps workflow, widened to a peer group

Same as above at a wider scale — list the peer set explicitly rather than asking Claude to define the sector.

Here is [metric, e.g. revenue growth] for [fiscal period] for these [N] companies in [sector/industry]: [list]. Rank them from highest to lowest, and note which ones are outliers more than one standard deviation from the group average.

4. Fair-value / DCF estimate

Closest official equivalent: DCF modeling (/dcf)

A DCF is only as good as its assumptions. Make Claude state them separately from the number, per the how-to steps below.

Using the following figures for [Company Name][free cash flow for the last N years] — build a simple discounted cash flow estimate for [valuation date]. List every assumption you use (growth rate, discount rate, terminal value method) in a separate section before the final estimate, and show the estimate as a range, not a single number.

5. Growth-rate comparison

Closest official equivalent: 3-statement financials workflow

Useful for spotting whether growth is accelerating or decelerating across several periods, not just two.

For [Company Name], using the [revenue and EPS] figures below for [the last 5 fiscal quarters], calculate the sequential and year-over-year growth rate for each period and tell me whether growth is accelerating, decelerating or flat. Show the calculation, not just the conclusion.

6. Idea screening

No direct official plugin equivalent — this is general reasoning, not a modeling workflow

Unlike the five above, Anthropic's plugin set doesn't have a named "screener" building block — this one is just structured brainstorming, and should be labeled as such to yourself.

I'm screening for companies with [criteria, e.g. positive free cash flow and revenue growth above 10%] in [sector]. From the list of candidates I paste below, tell me which ones meet each criterion individually and which fail, based only on the figures I provide — don't add companies I didn't list.
For Korean-market tickers: the regular Claude app has no live KRX feed and no built-in connector to Korea's disclosure system. To fill in the template above, pull the filing yourself from the Financial Supervisory Service's OpenDART portal, or export figures from a broker feed such as Korea Investment & Securities' Open Trading API, then paste the relevant numbers into the prompt. Neither service is affiliated with this site or with Anthropic.

How to build the prompt, step by step

  1. Name the company precisely — full name, ticker and exchange where relevant. A vague reference makes Claude guess which entity you mean.
  2. Pick exactly one metric — revenue growth, operating margin, free cash flow, EPS. One number per prompt, not a general opinion.
  3. Set the exact period — fiscal quarter or year, and year-over-year or quarter-over-quarter. Claude can't infer "recent" from a filing calendar it can't see.
  4. Supply the source data yourself — paste or upload the filing, transcript or export. The analysis is only as current as what you give it.
  5. Specify the output structure — ask for a table or fixed labeled sections so the answer is easy to check against the source.
  6. Ask for assumptions to be listed separately — especially for anything forward-looking like a DCF estimate, so you can challenge each input on its own.

What this is not

These are prompt-writing templates, not investment advice and not a signal service. Claude's output is not personalized financial advice, is not guaranteed to be accurate or current, and should never be the only input into a buy or sell decision — verify every figure against the primary filing or your exchange's own data before acting on it.

Frequently asked questions

Did Anthropic publish an official list of 6 Claude stock-analysis prompts?

No. That framing circulates on social media but there is no such document from Anthropic. What Anthropic does publish officially is a prompting-strategies guide for financial analysis on claude.com, plus an open-source Financial Services plugin set (github.com/anthropics/financial-services) built for Claude Cowork and Claude Code that runs comparable-company analysis, DCF models, LBO models and earnings-transcript analysis through slash commands such as /comps, /dcf and /earnings.

Is Anthropic's Financial Services plugin built for individual investors?

Not primarily. It is aimed at finance professionals inside Claude for Enterprise or Claude Cowork, and its deepest features rely on paid institutional data connectors such as FactSet, S&P Global, Moody's and PitchBook. An individual investor using the regular Claude app can still borrow its underlying prompt structure, but has to supply the source data manually rather than pull it from those connectors.

Can Claude pull live stock prices or Korean DART filings on its own?

Not in the standard consumer app. Claude does not have a built-in live market-data or DART connector for individual users. For Korean-market names, you generally retrieve the filing or figures yourself — for example from the Financial Supervisory Service's OpenDART portal or a broker's API such as Korea Investment & Securities' Open Trading API — and paste or upload the relevant text into the chat before asking Claude to analyze it.

Are these prompts investment advice?

No. These are prompt-writing templates for organizing your own research questions. Claude's output is not personalized investment advice, is not guaranteed to be accurate or current, and should not be the sole basis for a buy or sell decision. Always verify figures against the primary filing or exchange data.

Keep reading

This guide is for general information only and is not investment, financial, tax or legal advice. AI assistant features, pricing and official plugin availability change frequently — always verify current details on Anthropic's own site and repository before relying on them. Figures you generate with Claude should be checked against the primary source (company filing or exchange data) before you rely on them for any decision. We do not guarantee any specific output quality or accuracy.

More free guides from our network

Independent, ad-free buying guides and tools across related topics.

SaaS HubSaaS tools comparedWeb Hosting GuideHosting compared by use caseEmail Marketing SoftwareEmail platforms compared