Guides LLMs and Agents

LLMs and Agents

Learn how to leverage Large Language Models (LLMs) and AI Agents to build powerful crypto applications using Guavy.

Generating Crypto Applications

You can easily generate your own crypto applications using your favorite LLM (like ChatGPT, Claude, or Gemini) combined with our specialized context file.

We provide a comprehensive llms.txt file that contains the necessary context and API specifications for Guavy. By providing this file to your LLM, you can instruct it to write code that interacts with our API correctly.

Our context file is available at: https://data.guavy.com/llms.txt

Simply copy the contents of the llms.txt file or provide the URL to your LLM, and ask it to build a crypto dashboard, trading bot, or analysis tool using the Guavy API.

Example Prompts

Here are some example prompts you can use with your LLM to build applications with Guavy.

Crypto News & Sentiment Dashboard

Build a production-quality, full-stack Next.js application that functions as a modern cryptocurrency news, sentiment, and trading signals dashboard.

# Crypto News & Sentiment Dashboard (Guavy API)

Build a **production-quality, full-stack Next.js application** that functions as a modern cryptocurrency news, sentiment, and trading signals dashboard powered by the **Guavy API**.

**API Reference:**  
https://data.guavy.com/llms.txt

---

## Design Direction (High Priority)

Take strong visual inspiration from **popular trading apps**.

### Color & Theme
- Primary palette:
  - Black
  - Dark gray
  - Light gray
  - White
- Accent colors:
  - **Green** → positive price movement / bullish sentiment
  - **Red** → negative price movement / bearish sentiment
- Accents should be:
  - Subtle
  - Data-driven
  - Tasteful (no neon or overuse)

### Overall Feel
- Clean and modern
- Calm and professional
- High-signal / low-noise
- Finance-grade UI
- Optimized for readability and fast scanning

Use spacing, contrast, and typography hierarchy intentionally.  
This should feel like a serious crypto intelligence product, not a meme or hype-driven trading site.

---

## Tech Stack & Architecture

### Frameworks
- **Next.js (latest App Router)**
- **TypeScript**

### Styling
- **TailwindCSS** for core layout and styling
- **daisyUI** for UI components and theming
- Create a custom daisyUI theme matching the palette above
- Dark mode by default with a **dark/light mode toggle** via daisyUI

---

## Core Features

### 1. Dashboard / Home (`/`)
- Display a **live-updating crypto news feed** using the Guavy API
- Each news card should include:
  - Headline
  - Source
  - Published timestamp
  - Related coin(s)
  - Sentiment indicator (bullish / neutral / bearish when available)
- Visual sentiment cues:
  - Red / green accents
  - Badges, arrows, or minimal indicators (avoid noisy charts)
- Users can filter the feed by:
  - Selected coins
  - Sentiment (optional)
- Data updates via:
  - Server Actions + revalidation **or**
  - SWR

---

### 2. Following Coins
- Users can select which coins they want to follow
- Followed coins affect:
  - News feed filtering
  - Default dashboard view
- Store preferences using:
  - `localStorage` or browser storage (no authentication required)
- Provide a clean coin selector UI:
  - Dropdown, modal, or searchable list (using daisyUI components)

---

### 3. News Detail Page (`/news/[id]`)
- Clicking a news item opens a dedicated article page
- Page should display:
  - Full article content (fetched from Guavy API)
  - Article metadata (source, published time)
  - Coins mentioned in the article
- For each mentioned coin:
  - Sentiment analysis
  - Trading signals (if available)
  - Visual indicators for numeric sentiment values
    - Bars, gauges, or minimal micro-charts

---

## API Integration Rules (Strict)

- **All Guavy API calls must occur on the server only**
  - Use Route Handlers or Server Actions
  - Do NOT call the API directly from client components
- Centralize API logic in:
  - `lib/api.ts`
- Include helper utilities to:
  - Normalize API responses
  - Map sentiment values to UI states
- Gracefully handle:
  - Loading states (daisyUI skeletons)
  - Error states (daisyUI alerts)
  - Empty states (clear user messaging)

---

## Performance & Code Quality

- Use:
  - Next.js built-in caching or SWR for minimal caching
- Optimize for:
  - Fast initial load
  - Smooth client-side interactions
- Maintain clean structure:
  - `app/`
  - `components/`
  - `lib/`
  - `types/`
- Use reusable, well-typed components
- Clean separation between:
  - Server logic
  - Client interactivity
  - Presentation/UI

---

## Accessibility & UX

- Use accessible daisyUI components by default
- Ensure:
  - Keyboard navigation
  - Proper ARIA roles
  - Sufficient color contrast
- UX goals:
  - Fast
  - Predictable
  - Professional
  - Information-dense without being overwhelming

---

## Goal

The final application should feel like:

> **A clean, professional crypto intelligence dashboard inspired by popular trading apps — combining news, sentiment, and signals into a fast, trustworthy experience.**

Was this helpful?