Context Engine

The core decision system that queries the Reputation Oracle, evaluates transaction signals, and recommends the appropriate privacy level with a plain-language justification.

The Context Engine is the core decision system. It accepts a transaction request and outputs a recommended privacy level along with a plain-language justification. The decision is produced from a combination of deterministic rules and learned models.

Deterministic rules cover unambiguous cases such as known-public recipients or trivial amounts. Learned models cover ambiguous cases where context and pattern recognition produce better outcomes than rules alone. The user can always override the Engine's recommendation — and those overrides feed back into the model for continuous improvement.

Signal inputs (evaluated per transaction)

• Recipient Reputation Score: Queried from the Reputation Oracle — how exposed, risky, or linked is the destination address?
• Transaction Value vs. History: Is this a typical send or an outlier? A sudden $50,000 transfer triggers different logic than a recurring $50 payment.
• User Intent: Optional tags provided by the user (e.g., "rent payment", "business settlement").
• Cross-Chain Doxx Exposure: Has the user's address been linked to off-chain identity on any supported chain?
• User Privacy Score: The wallet's overall privacy hygiene rating, influencing how aggressively to protect the user.