Skip to main content

​
Introduction

An Agent App is an autonomous AI agent created using the BunnyOS SDK and run on the BunnyOS App Platform. An Agent App encapsulates a specific autonomous behavior or strategy. It continuously observes the DeFi environment, evaluates conditions, and executes on-chain actions within predefined safety boundaries. Unlike traditional bots or scripts, an Agent App is designed to operate as a long-lived system. It maintains persistent state, adapts to changing conditions, and functions independently without constant human intervention.

​
What an Agent App Contains

An Agent App is composed of three core elements: Strategy and Decision Logic This defines how the agent evaluates information and decides what to do. It includes:
  • Rules or models for interpreting market and protocol data
  • Conditions under which actions should be taken or avoided
  • Decision logic for selecting between multiple possible actions
This logic is defined using the BunnyOS SDK. Operational Constraints Every Agent App operates within explicit boundaries. These constraints include:
  • Approved protocols and actions
  • Spend limits and exposure limits
  • Risk controls and safety checks
Constraints are enforced by the BunnyOS SDK and validated again at execution time by the BunnyOS App Platform. Long-Lived State An Agent App maintains awareness of its past actions and current context. This includes:
  • Current positions and allocations
  • Historical executions and outcomes
  • Pending or in-progress actions
State is persisted by the BunnyOS App Platform, allowing the Agent App to behave consistently over time rather than restarting from scratch.

​
How Agent Apps Operate

Once deployed, an Agent App runs continuously on the BunnyOS App Platform.
  1. The platform monitors blockchains, protocols, and data sources
  2. A relevant trigger occurs
  3. The Agent App is invoked with current state and data
  4. The app evaluates conditions using its SDK-defined logic
  5. If action is warranted, the app constructs a safe on-chain action
  6. Approved actions are executed on-chain
  7. State is updated and persisted
This loop repeats indefinitely, allowing the Agent App to operate autonomously over long time horizons.

​
Examples of Agent Apps

Agent Apps can represent a wide range of autonomous DeFi behaviors, including:
  • Portfolio rebalancing agents
  • Yield optimization agents
  • Risk monitoring and hedging agents
  • Liquidity management agents
  • Protocol-specific strategy agents
Each Agent App is a standalone autonomous economic actor that interacts directly with DeFi protocols.

​
Why Agent Apps Matter

Agent Apps transform DeFi automation from reactive scripts into persistent intelligent systems. By combining autonomous reasoning, continuous execution, and enforced safety constraints, Agent Apps enable more sophisticated, reliable, and scalable financial automation.