Risk before execution.

Core principles

APENT is designed to fail closed. Where a mandatory asset, data, liquidity, contract, network, or transaction validation cannot be completed, new exposure should not be created. Missing an opportunity is preferable to knowingly bypassing a critical control.

Execution authority

The strategy layer proposes actions. The risk layer evaluates whether those actions are permitted. The execution layer should only sign and submit transactions that match the approved action and active network policy. Autonomous execution authority should be narrower than administrative authority.

Market and liquidity risk

Autonomous trading is exposed to price movement, volatility, spread, slippage, insufficient liquidity, failed exits, and rapid changes in market structure. Position sizing and exposure limits should reflect observable liquidity and expected transaction conditions rather than target returns.

Smart-contract and asset risk

APENT should restrict interaction to approved assets and execution contracts. Contract address, chain ID, permitted function, amount, allowance, deadline, and expected output should be validated before signing. Unknown contracts are not treated as acceptable because they share a familiar ticker or name.

Market-data and oracle risk

Incorrect, stale, unavailable, or inconsistent data can produce invalid decisions. Data age and source status should be evaluated before execution. Where a strategy depends on multiple sources, material divergence should trigger a restrictive response rather than an invented midpoint.

Operational and network risk

APENT depends on blockchain access, transaction propagation, signing systems, monitoring, and external data services. Service interruption can delay observation, execution, or confirmation. Critical system failures should block new exposure and preserve the ability to reduce existing risk where possible.

Operating states

APENT's operating model supports distinct states such as Active, Reduce Only, Paused, and Emergency. The autonomous strategy should not be able to override a restrictive state or increase its own risk permissions.