> For the complete documentation index, see [llms.txt](https://cattoverse-1.gitbook.io/cattoverse/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://cattoverse-1.gitbook.io/cattoverse/introduction/agent-system-architecture.md).

# Agent System Architecture

Catto is built on a multi-layer agent architecture. Each layer plays a specific role on the path from market state to user-controlled automated outcomes.

#### **👁️ Perception Layer**

Continuously ingests and scores signals from on-chain data, wallet tracking, social sentiment, market structure, protocol metadata, and news catalysts, ranked by relevance against your portfolio context.

#### **🧠 Reasoning Layer**

Interprets the signal environment to generate actionable opportunity assessments, risk condition evaluations, strategy-fit scoring, and recommended execution actions with confidence levels. This layer is the intelligence core of the agent, deciding what matters, what to surface, and what to act on.

#### **⚡ Execution Layer**

Translates decisions into blockchain-correct transaction objects, including asset resolution, route construction, fee estimation, approval management, payload assembly, and confirmation tracking. Operates only within your authorized parameters and cannot self-expand its permission boundaries.

#### **🎚️ Autonomy Control Layer**

You control how much the agent can act independently:

🔔 Notify only, Catto alerts you, you execute manually 

📋 Prepare and confirm, Catto prepares the transaction, you sign to authorize 

🤖 Conditional auto-execute, Catto executes within pre-authorized strategy rules 

⏰ Scheduled execution, Catto runs defined tasks on a time-based schedule

At no autonomy level does Catto hold private keys. All execution flows through wallet-native signing.