Native MCP & llms.txt Support

Market sentiment your agent can call directly.

Six-dimension sentiment, trade signals, and curated news, delivered over a clean REST API and native MCP, so the agent reasoning about a trade reads the signal directly. Crypto, forex, commodities, and stocks are live today. Works with Claude, ChatGPT, Gemini, Grok, and any LLM out of the box.

Get free API key View llms.txt

300+ assets tracked

Trend history & backtests

Scored sentiment & signals

MCP native support

Free sandbox forever

// The problem

Market intelligence wasn't built for agents

The incumbents have sentiment. It just lives where an agent can't reach it: inside a terminal or a closed model, priced for the institution and built for a human watching a screen.

Locked inside a terminal

Legacy market intelligence is delivered to an analyst staring at a screen. Agentic systems don't watch screens. Terminals were built for a human at a desk; Guavy was built for the agent making the decision.

Raw data has no scored context

A price tells your agent nothing on its own. Is sentiment rising or falling? Is the crowd speculating or convinced? Guavy scores the mood across six dimensions, so the signal is structured, not noise.

No standard agent integration

Most providers have no MCP server, no llms.txt, and no concept of agent-to-API access, so you write glue code. Guavy is API-first and MCP-native from day one.

Guavy is the layer agents call

Six-dimension sentiment, scored news, and trade signals over REST and native MCP. Institutional sentiment, without the terminal. Your agent is live in minutes.

// Use cases

What builders are creating

From solo developers to institutional quant teams, here's how Guavy's API powers real agentic workflows.

01 / Sentiment Monitor

Sentiment-triggered alert agent

An agent that polls sentiment history across a portfolio of assets and fires alerts when sentiment crosses defined thresholds. No manual monitoring required.

get-sentiment-history get-recent-briefs

02 / Trading Bot

Signal-driven autonomous trader

Feed Guavy's buy/sell/hold signals and backtest data into an LLM-powered trading agent. Let it reason about when to enter and exit positions using AI-validated trend data.

get-current-action get-recommendations get-backtest-summary

03 / Research Agent

Daily market intelligence brief

An agent that pulls the AI market summary, top movers, and instrument-specific analysis each morning and composes a concise brief delivered to Slack, email, or a dashboard.

get-market-summary get-instrument-analysis get-trend-history

// Quickstart

Live in under 5 minutes

No SDKs to install. Standard REST with Bearer token auth. Works with any HTTP client.

1

Get a free API key

Sign up at guavy.com. Your key is ready instantly.
2

Start making API calls

Use standard REST endpoints from any HTTP client. For AI integrations, paste https://guavy.com/llms.txt into Claude, GPT, or any LLM, and it auto-discovers every endpoint.
3

Or call the REST API directly

Standard GET endpoints. JSON responses. Add your Bearer token to the Authorization header and you're calling the API in seconds.
4

Connect via MCP (optional)

For native agent tool-use, use Guavy's MCP server to give your agent first-class access to sentiment, signals, and news as callable tools.

python

# Get live Bitcoin sentiment for your agent
import requests

API_KEY = "YOUR_GUAVY_API_KEY"
BASE = "https://guavy.com/api/v1"

headers = {
  "Authorization": f"Bearer {API_KEY}"
}

# Fetch today's BTC sentiment
r = requests.get(
  f"{BASE}/sentiment/get-sentiment-history/btc",
  headers=headers,
  params={"limit": 1}
)

today = r.json()["sentiment"][0]
print(f"BTC positive: {today['positive']}")
print(f"BTC negative: {today['negative']}")

# Get trade signal
signal = requests.get(
  f"{BASE}/trades/get-current-action/btc/aggressive",
  headers=headers
).json()

print(f"Signal: {signal['action']}")
# → Signal: Buy

// Integrations

Works with every LLM and framework

Native llms.txt support means any LLM can understand and call the Guavy API without custom instructions.

MCP Native

llms.txt

Claude (Anthropic)

ChatGPT

Gemini

Grok

LangChain

LlamaIndex

CrewAI

AutoGen

Any REST client

// For AI agents

Your LLM already knows how to use Guavy

No prompt engineering. No custom function definitions. No glue code. Just point your agent at one URL.

llms.txt: Machine-Readable API Spec

guavy.com/llms.txt

The complete Guavy API specification (every endpoint, parameter, and response field) formatted for direct LLM consumption. Point your agent at this URL and it will understand exactly what to call and how, with no additional setup required.

Claude Attach https://guavy.com/llms.txt as a document or paste the URL. Claude reads the spec and writes correct API calls immediately.

ChatGPT / Gemini / Grok Include the llms.txt URL in your system prompt or as context. The model will use it to generate valid requests without any custom tooling.

Any Agent Framework LangChain, CrewAI, AutoGen: fetch llms.txt at runtime to give your agent a live, always-current view of the API surface.

View llms.txt

MCP

Model Context Protocol

Native tool-use for agents

For agents that use tool-calling natively, Guavy's MCP server exposes sentiment, signals, and news as first-class callable tools. No REST wrappers needed.

MCP Setup Guide

Full API Reference

Prefer to build with the REST API directly? The complete docs cover every endpoint with example requests, response schemas, and code samples.

Read the docs

// Pricing

Start free. Scale as you build.

All prices in USD.

Sandbox

^$0

/ month · free forever

Every live market (REST & MCP)
Sentiment & news
Unlimited Wire articles
3,000 tokens / month per market
Personal use

Get started free

Questions from AI builders

How do I connect Guavy to Claude or another LLM?

The easiest way is to give Claude or any LLM the URL https://guavy.com/llms.txt. It contains the complete API spec in a format LLMs understand natively, so your agent knows every endpoint, parameter, and response format with no extra setup. For MCP-native integration, use the Guavy MCP server so your agent calls market data as first-class tools.

Does Guavy support the Model Context Protocol (MCP)?

Yes. Guavy has a native MCP server at https://guavy.com/mcp, MCP-native from the start, with the same tool surface across every market, so an agent you build on one market's tools extends to the next without re-architecting. Your agent discovers and calls every Guavy tool with no glue code. Works with Claude Desktop, Cursor, VS Code, Grok, Gemini CLI, and LangChain. Step-by-step setup guide

How fresh is the data?

Sentiment scores, news briefs, and trade signals are updated continuously throughout the day. Price and market cap data reflects current market conditions. The get-market-summary and get-instrument-analysis endpoints are refreshed daily with AI-generated summaries of current conditions.

Can I use Guavy data in a commercial product or SaaS app?

The self-serve plans (Sandbox, Quant, and Pro) are licensed for personal use. Commercial use in apps, funds, or agents, plus white-label data and redistribution, is covered by a Business license: Commercial starts at $499 / month, and Enterprise is priced to your usage and scale. See business pricing or talk to our sales team to scope it.

Which markets does Guavy cover, and how?

Crypto, forex, commodities, and stocks are live today. Every market is scored across the same six dimensions and available via both REST and MCP. Crypto alone spans 300+ assets; use the list-symbols endpoint to pull the full list programmatically.

Is there a rate limit I need to worry about?

The API has a limit of 10 simultaneous connections. Monthly token allowances depend on your plan: 3,000 for Sandbox, 250,000 for Quant, and 1,000,000 for Pro. For large data syncs, use pagination parameters (limit and skip) to batch requests efficiently.

// Get started

Your agent is missing
market sentiment

Free sandbox forever. Your first call in minutes.

Get free API key See Recipes Read the docs