LogoQUANTUMSTACK

Agent Communication Guide

This guide explains how to communicate effectively with your AI agent, understand its capabilities and integrate it with the QuantumStack ecosystem.

What the Agent Is

The agent is an autonomous AI system that executes actions according to rules defined in your NFT.

It works as an intelligent assistant that can execute transactions, monitor markets, manage wallets and more, always respecting the rules defined in the NFT that governs it.

Fundamental Principles

  • Sovereignty: The agent executes only what the NFT allows
  • Decentralization: Does not depend on a centralized executor
  • Transparency: All actions are verifiable on-chain
  • Evolution: The agent evolves according to NFT rules

How the Agent Works

The agent operates through a secure, rule-governed flow: it receives your request, validates it against the permissions embedded in your NFT, and either executes the action or explains why it cannot.

Command Types

1. Query Commands

Commands that only query information, without executing transactions:

  • "What is my balance?" - Query wallet balance
  • "Show my recent transactions" - List history
  • "What is the current price of ETH?" - Query prices
  • "What are my current rules?" - Shows NFT rules

Example: You ask "What is my balance?" and the agent queries the smart wallet, returns the current balance and may show recent transaction history.

2. Configuration Commands

Commands that configure agent behavior:

  • "Monitor BTC price" - Configure monitoring
  • "Alert me when ETH rises 5%" - Define alerts
  • "Enable staking module" - Enable features
  • "Switch personality to conservative" - Adjust behavior

Example: You say "Monitor BTC price and notify me if it drops more than 10%". The agent configures monitoring and notifies you when the condition is met.

3. Action Commands

Commands that result in on-chain transactions (require validation):

  • "Stake 50% of my funds" - Execute staking
  • "Buy 0.1 ETH when price drops 5%" - Configure trading
  • "Send 0.05 ETH to [address]" - Transfer (if allowed)
  • "Rebalance my wallet" - Rebalancing

Example: You say "Stake 50% of my funds". The agent checks if the amount is within NFT limits, creates a transaction proposal, and the wallet validates before executing on-chain.

Internally, each agent is isolated by its NFT identity — rules, context and execution are scoped exclusively to the owner's token.

Communication Best Practices

Be Specific

Specific commands are more efficient:

  • Good: "Monitor ETH price and buy 0.1 ETH when it drops 5%"
  • Avoid: "Monitor things and do something when it changes"

Respect the Limits

The agent always respects the limits defined in the NFT:

  • Maximum values per transaction
  • Allowed destinations
  • Operation frequency
  • Enabled action types

Use Natural Language

The agent understands natural language:

  • You do not need to use specific technical commands
  • You can ask questions as in a normal conversation
  • The agent interprets intent and context
  • It may ask for clarification if needed

Practical Examples

Example 1: Balance Query

Command: "What is my current balance?"

What happens:

  1. Agent receives command
  2. Queries smart wallet linked to NFT
  3. Returns current balance and optionally recent history
  4. No transaction is executed

Example 2: Price Monitoring

Command: "Monitor BTC price and notify me if it rises more than 10%"

What happens:

  1. Agent configures price monitoring
  2. Validates that monitoring is allowed in NFT rules
  3. Starts continuous monitoring
  4. When condition is met, notifies you
  5. No transaction is executed automatically

Example 3: Automatic Staking

Command: "Stake 50% of my funds when there is an opportunity"

What happens:

  1. Agent calculates 50% of current balance
  2. Validates if amount is within NFT limits
  3. Searches for available staking opportunities
  4. Creates staking transaction proposal
  5. Wallet validates proposal against NFT rules
  6. If approved, executes on-chain
  7. Agent confirms execution

Every action goes through enterprise-grade validation before execution. Critical rules are enforced on-chain — the system rejects unauthorized actions even in edge cases, and all operations are permanently recorded.

Troubleshooting

Agent not responding

  • Check if the NFT is active and linked
  • Confirm that the agent is initialized
  • Check connection to decentralized storage
  • Try restarting the agent if needed

Action rejected

  • Check if the action is within NFT limits
  • Confirm that the destination is allowed
  • Check if there is sufficient balance
  • Review NFT rules to understand the rejection

Command not understood

  • Try rephrasing the command more clearly
  • Use more specific language
  • Ask the agent what it understood
  • Consult examples of valid commands

Next Steps

To learn more about the agent and QuantumStack:

Last updated: January 2025
Document version: 1.0.0