A Ready-Made Suite of Financial Tools for AI Agents

If you've ever asked Claude or GPT-4 to calculate WACC, build a DCF model, or assess how an earnings report impacts stock prices, you've likely run into the same problem. The model either brazenly makes up numbers or spits out generic textbook economics. To get real analytics, neural networks need access to proper data sources and ready-made calculation algorithms.

Developers have released a finance-skills repository on GitHub. It's a collection of plugins and skill extensions built on the open Agent Skills standard. It turns Claude Code and other autonomous agents into a financial assistant.

What's Inside the Repository

The project is built not as a monolithic service, but as a set of separate modules. Each skill group handles its own layer of data processing.

The first plugin, finance-market-analysis, takes care of routine company valuation tasks. Instead of manually gathering reports through yfinance, the model can request ready-made Bull/Base/Bear scenarios, calculate stock price sensitivity to WACC, analyze ETF premium or discount to NAV, and check trend patterns like VCP using the Minervini strategy.

The finance-social-readers block handles reading social networks and specialized platforms. Through opencli and tdl utilities, the agent can peek into Telegram channels, read threads on Twitter, Reddit, LinkedIn, and even pull startup data from Y Combinator. Access works in read-only mode, so there's no risk of accidentally publishing anything.

The third block, finance-data-providers, connects external specialized services. This includes integrations with Fintel for tracking short interest and insider trades, sentiment gathering from Adanos, as well as reading order books and funding from the Hyperliquid crypto exchange. There's also a tool for working with the desktop version of TradingView via Chrome DevTools Protocol.

There are also less obvious things. For example, finance-startup-tools breaks down startup analytics from three perspectives: venture investor, job seeker, and founder. And the finance-ui-tools plugin leverages generative UI in Claude to display interactive charts and widgets right in the chat.

How It's Built Technically

All skills are written according to the Agent Skills specification. They can be connected not only to Claude Code, but also to any agents that support this format.

Installation in Claude Code is done with a single terminal command:

npx plugins add himself65/finance-skills

If you don't need the entire package, you can easily install only specific modules. For example, if you're interested in market analysis and the TradingView order book, just install the individual plugins:

npx plugins add himself65/finance-skills --plugin finance-market-analysis
npx plugins add himself65/finance-skills --plugin finance-data-providers

For third-party frameworks and agents, a universal CLI is used:

npx skills add himself65/finance-skills -a <agent-name>

Under the hood, many readers rely on the opencli utility. It handles parsing web pages and converting responses into an LLM-friendly format. For working with TradingView charts and options, the plugin connects to the running application via the debug port, reads chart states, options board data, and can take screenshots.

Where This Really Comes in Handy

Instead of abstract discussions, let's look at three practical scenarios where these skills will save time.

Preparing for a quarterly earnings report. Usually before a report comes out, you have to open several tabs: analyst consensus forecasts, history of past EPS surprises, short interest dynamics on Fintel, and discussions on Twitter. With the skills connected, you just give the agent a command to compile a summary for a ticker. The model will itself go into Fintel, pull data from yfinance, check recent tweets, and deliver a structured briefing.

Monitoring arbitrage on the crypto market. The Hyperliquid plugin reads funding rates on perpetual futures and compares them across exchange pools. The agent can monitor anomalies in the background and highlight inter-exchange spreads.

Evaluating startups and SaaS companies. If you're evaluating a tech business, the saas-valuation-compression plugin will break down ARR multiples and compare them with competitors. The startup-analysis tool at the same time will analyze business model risks, generating a report from both an investor's and a potential employee's perspective.

Is It Worth Installing

The project will be useful for developers building their own finance-oriented agents and traders using Claude Code as a daily terminal. You can't build a full trading system on this without customization. The author immediately warns in the disclaimer: the project was created for educational purposes.

The code is transparent, the MIT license allows free customization of skills for your needs, and the authors even included a separate finance-skill-creator plugin for creating and evaluating the quality of your own skills. If you've been wanting to automate financial analytics collection via LLM, this is a great starting point.