Lylacore SDK

Use the validateMind function from lylacore/sdk/mind-loader, passing your Mind data object and the loaded schema. It returns a result object with a valid flag and an errors array listing any schema violations.

Load your Minds with loadMindsFromDirectory, then call exportSoulPackage with the minds array and metadata such as author, description, and tags. The resulting package object can be serialized to JSON and saved to a file.

A Mind requires a name, a Unicode symbol of at most four characters, and a role describing its primary function. Optional fields include traits, triggers with context patterns and activation weights, and coordination rules for multi-Mind interaction.

Yes, importSoulPackage accepts a selectMinds option containing an array of Mind names. Only the matching Minds are returned along with the package metadata, letting you load a subset of a larger package.

The learnPattern function analyzes each user message and agent response pair with context like language and topic, producing style metadata. The applyStyle function then rewrites outgoing messages to match the learned language, formality, and tone.

Yes, the documentation recommends encrypting sensitive identity data for production deployments using a soul-crypto module. Encrypted Mind data should be decrypted before validation against the Lylacore schema.