Agent Interface

The Agent interface provides multiple ways to understand, interact with, and monitor how an agent manages a vault.

The interface is composed of four main sections:

• Agent Overview

• Chat

• Action Logs

• Terminal

Each section answers a different class of questions about the agent and the vault it manages.

---

Agent Overview

The Agent Overview presents a high-level snapshot of the agent’s current state and configuration.

It allows you to quickly understand:

• Whether the auto-farm agent is enabled and allowed to run transactions

• The agent’s current goal and risk profile

• What the agent is actively managing

• When the agent is expected to run next

This view is intended for quick monitoring and status checks, without requiring deeper inspection of logic or execution details.

---

Chat

The Chat tab provides a conversational interface for interacting with the agent using natural language.

The Chat interface is context-aware and uses the current vault and agent state to provide relevant responses. It allows you to ask questions and issue requests related to:

• Vault assets, positions, and performance

• Vault-related data and metrics

• Transaction-related actions

• Liquidity pool data

• Liquidity provisioning (LP) strategies

Depending on the request, the agent may explain information, prepare an action for execution, or guide you through the next steps.

Chat is designed for exploration, understanding, and interaction, making it easier to work with vaults without navigating complex settings or documentation.

---

Action Logs

Action Logs record the actions the agent attempts to perform while managing a vault.

When the agent determines that an action is required, it creates an action plan and attempts to execute it. Action Logs capture the outcome of those attempts.

Each Action Log reflects:

• The action the agent attempted

• The execution process

• The final result, such as

  • Success: all actions in the plan were executed successfully

  • Partial Success: the action plan contains multiple steps, and some of them failed while others succeeded

  • Failed: the action could not be executed, with error details provided

Action Logs provide visibility into what the agent tried to do and what ultimately happened, supporting transparency and auditability.

---

Terminal

The Terminal provides a low-level view into the agent’s internal decision context.

It exposes the structured information the agent uses when forming decisions, including:

• The decision prompt and inputs

• The vault state and assumptions at the time of evaluation

• The context used to generate an action plan

The Terminal is intended for advanced users, strategy designers, and debugging workflows where deeper inspection is required.

---

How These Sections Work Together

Each section of the Agent interface answers a different question:

• Agent Overview — What is the agent doing right now?

• Chat — How can I interact with or understand the agent using natural language?

• Action Logs — What actions were attempted and what were the outcomes?

• Terminal — How did the agent arrive at its decisions?

Together, they provide a complete and transparent view of the agent lifecycle—from configuration and reasoning to execution and results.

---

Summary

The Agent interface is designed to make automated vault management:

• Understandable through natural language

• Observable through structured logs

• Auditable through execution records

• Flexible for both simple and advanced use cases

By combining overview, interaction, logging, and diagnostics, users gain confidence and control over how agents manage their vaults.