Agent Definition Schema

To safely onboard custom agents into the system, Agentify requires each agent to follow a clear definition schema.

What’s in the schema?

  • Manifest: Name, description, version, creator.

  • Capabilities: What actions the agent can perform (like "swap", "lend", "borrow").

  • Permissions: Which smart contracts, tokens, or chains it can touch (used for sandboxing).

  • AI Behavior Spec: Defines how the agent interprets user prompts and maps them to actions.

    • Could be done via:

      • Prompt templates (e.g., "When user says X, do Y")

      • State machine logic (workflow-based behavior)

      • Scripting or code-based decisions

This format ensures interoperability, security, and predictable behavior.

PreviousSDK and APIsNextAI Playground (Sandbox)

Last updated