> ## Documentation Index
> Fetch the complete documentation index at: https://docs.guppyapi.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Core Concepts

> Understand the fundamental building blocks of Guppy

### Intent

An **Intent** represents a user’s desired outcome over time. Unlike a traditional API call that executes immediately, an intent is a declaration of a goal.

Examples:

* "Buy PS5 Pro if it restocks at Target for \< \$800. Max 1."
* "Reorder coffee beans when inventory \< 10. Net‑30 only."
* "Reserve a table Friday or Saturday between 7–9pm; cancel if not confirmed by Thursday."

Intents are **stateful**, **long‑lived**, and **conditional**.

***

### Guardrails

Guardrails constrain what Guppy is allowed to do. They ensure that execution stays within the bounds of user intent.

Common guardrails:

* Max price / quantity
* Allowed merchants
* Payment method constraints
* Time windows / expiration
* Retry & fallback policies

***

### Plan (Internal)

A **Plan** is Guppy’s internal state machine derived from an intent. Guppy analyzes the intent and guardrails to determine the best path for execution.

Developers do not directly manipulate plans; they observe progress via events.

***

### Execution Backends

Guppy routes execution across the most appropriate path:

* **Commerce APIs**: Modern, opt‑in stacks for direct execution.
* **UI‑based Web Execution**: Legacy sites and browser workflows where no API exists.
* **Fallbacks & Recovery**: Automatic retries, alternates, and refunds.

Developers never need to specify *how* execution happens. Guppy abstracts the complexity of the web.