Understanding Owner and Agent Authority for Trading Agents

Owner authority is the ultimate control over an account, able to approve any transaction, including withdrawals. Agent authority is a delegated permission set that the owner configures with specific scopes and limits. The distinction ensures that a trading agent can act without holding the power to move funds out of the account.

What is owner authority?

Owner authority resides in a single cryptographic key that the account holder safeguards. This key can sign any instruction the platform accepts, from placing orders to withdrawing assets. Because it is the only key that can approve withdrawals, the owner retains final say over capital movement. The downside is that loss or compromise of this key gives an attacker full control, so secure storage is essential.

What is agent authority?

Agent authority is created by the owner through signed policy documents that define what the agent may do. Typical policy fields include maximum order size, daily notional exposure, daily loss caps, and expiry timestamps. An agent key can only act within these bounds; it cannot sign a withdrawal intent. This limited scope reduces risk, but it does not eliminate operational or market risk inherent in the trading strategy.

How does owner authority protect withdrawals?

A withdrawal request must be signed by the owner key and include the exact intent (asset, amount, destination). Even if an agent key is compromised, it cannot initiate a withdrawal because it lacks the required signature. However, if the owner’s limits are too permissive, an attacker who gains the owner key could still move funds freely.

Can an agent key place trades without withdrawal rights?

Yes. A trade‑scoped agent key can submit orders that respect the owner‑defined limits. The platform enforces these limits at runtime, rejecting orders that exceed them. The agent may still generate losses up to the configured daily loss cap, and it cannot be stopped automatically if market conditions change dramatically; the owner must intervene.

How can you configure safe limits?

• ·Set a maximum order size that matches your risk tolerance.
• ·Define a daily notional ceiling to cap total exposure.
• ·Apply a daily loss limit to stop the agent after a predefined loss.
• ·Include an expiry timestamp so the agent’s authority ends automatically.
• ·Review and adjust limits regularly as market conditions evolve.

For detailed guidance on setting least‑privilege policies, see Understanding Least Privilege for an AI Trading Agent.

Never grant an agent the owner’s withdrawal key; keep that capability separate and tightly controlled.

The principle is reinforced in Why an AI Trading Agent Should Never Hold the Owner’s Withdrawal Key, which explains the security rationale behind separating trade and withdrawal authority.

A broader overview of autonomous agents can be found in What Is an Autonomous Trading Agent?.