AlphaGBM Skills: Options and Stock Intelligence Inside AI Coding Agents

What AlphaGBM Skills Does and Who It Is For

AlphaGBM Skills addresses a gap in AI coding environments: financial market research requires structured analysis methods that general-purpose AI assistants do not have natively. An analyst using Claude Code or Cursor to work on investment-related code would otherwise need to write Python scripts that call financial data APIs, parse the responses, and apply analysis logic. AlphaGBM Skills packages those methods as installable skills that the AI agent can invoke through natural-language requests.

The README describes the core value as "market research inside your AI workspace." The skills cover stock analysis, options strategy scoring, news impact interpretation, institutional research report breakdown, and investment review (comparing two snapshots of a portfolio). A user tells their AI agent to "Use AlphaGBM to research NVDA" and the skill handles the data retrieval and analysis presentation.

The project targets individual investors and developers building investment workflows, not enterprise trading desks. The README's verification boundary documentation and the explicit note that "research is not a return guarantee or an order to trade" frame this as a decision-support tool, not an execution system.

27 Skills Across Four Asset Categories

The catalog as of the last push on 2026-09-24 contains 27 skills organised into five core packages and 22 focused packages across four categories.

The five core skills are: - alphagbm-stock-research: fundamentals, sentiment, risk, and opportunity scores for a stock - alphagbm-options-research: candidate scores, capital requirements, and risk limitations for options strategies - alphagbm-news-impact: reported facts, affected assets, and impact inferences from news input - alphagbm-report-breakdown: institutional ratings, assumptions, and sources from research reports - alphagbm-investment-review: comparison of two user-supplied portfolio snapshots, run locally

The focused skills add nine stock-specific packages (including Dividend Strategy, Market Sentiment with VIX and fear indicators, ETF Strategy, Grid Plan, Dollar-Cost Averaging, Smart Money Tracking, and Momentum Following) and eleven options packages covering scoring, volatility, Greeks, strategy comparisons, payoff analysis, and risk methods.

Commodities and digital assets directories exist but are documented as reference packages that explain methods rather than exposing live API calls. The catalog at catalog/catalog.json defines the canonical package names, categories, and file paths. The README states the website imports a commit-pinned copy of this file.

Installing a Skill Into Claude Code or Cursor

Skills are installed using npx and the skills CLI. To see all available packages:

npx skills add AlphaGBM/skills --list

To install a single skill, pass the package name:

npx skills add AlphaGBM/skills --skill alphagbm-stock-research

To install a focused tool:

npx skills add AlphaGBM/skills --skill alphagbm-etf-strategy

The installer prompts for the target AI tool. After installation, the skill is available in that environment. Callable packages contain a self-contained Python 3.9+ runner; no separate AlphaGBM CLI installation is required.

An ALPHAGBM_API_KEY environment variable must be configured in the AI tool's environment. The key is created in the user's account at alphagbm.com/api-keys. The README warns: "Never paste a key into a conversation or commit it to a repository."

The Investment Review skill (alphagbm-investment-review) is the exception: it compares two portfolio snapshot files the user supplies locally and does not require an API call to AlphaGBM's servers.

Access Model: Free Installation, Account-Gated Research

Installation is free and requires only Node.js for npx. Account-backed research calls share the user's AlphaGBM website allowance and subscription rules. The README draws a careful line between what is free and what requires an account:

- Published research reads do not need a key but do not grant access to the private research archive. - Investment Review compares files locally with no API call. - Reference packages explain methods without an API. - Callable packages that retrieve live market data require a valid API key and consume the account's allowance.

Installation does not unlock Alpha Agent or grant additional quota beyond what the account plan covers. The README links to alphagbm.com/pricing for plan details.

The catalog marks each package's access and release status. A preview label does not mean production availability, and the README notes that "the catalogue preserves each package's access and release status; a preview is not a claim of verified production availability." Checking the catalog's current state before relying on a specific skill is the recommended practice.

Demo Fixtures Versus Live Responses

The repository includes a demo/ directory with example outputs for all current packages. The README is explicit that these are either synthetic output fixtures or source-based example requests: "Neither is presented as a captured live paid response."

This is a meaningful boundary. Analysts evaluating the tool must distinguish between the demo outputs (which show what a response looks like) and a live API response (which reflects current market data and the account's live analysis). The README provides a demo guide (demo/README.md) and a catalog index (demo/CATALOG.md) for navigating the examples.

The development commands show how contributors build and test the catalog:

python3 scripts/build_catalog.py
python3 scripts/build_catalog.py --check
python3 -m unittest discover -s scripts -p 'test_*.py'

And to verify the local catalog from the skills CLI perspective:

npx skills add . --list

The build script generates the catalog from skill definitions; the check flag validates it without writing. Unit tests are in the scripts/ directory.

Direct Financial API Access as the Alternative

The direct alternative is calling financial data APIs from within an AI coding session without an intermediary skill layer. Libraries like yfinance (Yahoo Finance's unofficial Python client), Alpha Vantage's official Python package, and Polygon.io's client are all usable from within Claude Code or Cursor by writing a few lines of code.

The key difference is effort and structure. A raw yfinance call returns data; the analyst or AI must then decide how to interpret it, what to calculate, and how to present the result. AlphaGBM's skills provide pre-structured analysis methods (scoring models, Greeks calculations, strategy comparisons) that the AI agent can invoke and explain without the analyst writing that logic. The trade-off is flexibility versus speed: raw API calls can be tailored to any calculation, while AlphaGBM skills are constrained to the methods the team has packaged.

The Repository also separates itself from the direct-API approach by the verification layer: runners check supported workflow contracts before applicable paid calls and the README states they "never replace failures with demos." A raw API call that fails returns an error the analyst must handle manually.

AlphaGBM/investment-masters is mentioned in the README as a related repository maintained separately, containing investor-inspired methods not duplicated here.

Editorial conclusion

AlphaGBM Skills suits developers and individual investors who work inside Claude Code or a compatible AI agent environment and want structured market-research methods available in their workspace without writing API integration code. It is not the right choice for teams that need deterministic, auditable backtesting or direct algorithmic trading integration: the README is clear that research output is not a return guarantee or a trade order. Before installing, confirm that your agent environment supports the skills invocation protocol, that your AlphaGBM account plan covers the analysis types you need, and that the live-data capabilities you intend to use are not labeled as reference packages in the current catalog.

Frequently asked questions

Official sources

1. AlphaGBM/skills on GitHub
2. Issues
3. License: MIT
4. Project website
5. README