Managing Gpt Email Generation

When to Use This Skill

Manage the GPT Email Generation (Sales Email Assistant) setup page. Reads/writes the EinsteinGPTSalesEmails master pref plus 8 related ECI/Sales-Signals/Coaching feature toggles. Also exposes a composite getState aggregator (28+ keys: pref values, feature-access flags, Hawking ML-platform tenant provisioning status, Conversation Mining provisioning status, CDP/Genie status, Signals2DC status, computed UI-state enum ENABLED/DISABLED/IN_PROGRESS/HAWKING_TENANT_ERROR) and a refresh action that kicks off Conversation Mining background work. All toggles return Boolean (resulting pref/perm value); some have value-add behavior agents must know about — sales-email enable atomically also flips OnlyUseBuilderPrompts so open-prompt is on by default; open-prompt's enabled=true means OnlyUseBuilderPrompts=false (semantic inversion).

Workflow

Read get-state first to discover current pref/perm/provisioning state and the computed UI-state enum (ENABLED/DISABLED/IN_PROGRESS/HAWKING_TENANT_ERROR). Toggle ordering: when enabling Sales Email Assistant via toggle-sales-email-pref, run toggle-open-prompt AFTER if you want explicit open-prompt control (enabling sales-email auto-flips OnlyUseBuilderPrompts to false in legacy-discovery-off orgs). For Conversation Mining: toggle-sales-conversation-mining-pref must run before toggle-sales-conversation-mining-apd360-perm and refresh-sales-mining (the perm and refresh sit on top of the feature pref). Each write returns Boolean (the resulting pref/perm value); follow with the paired verify-* step to read back via get-state and confirm. Note that toggle-sales-email-pref's enable path performs an atomic dual-pref bulk write (EinsteinGPTSalesEmails=true + OnlyUseBuilderPrompts=false); a single failure rolls back both. Hawking ML-platform tenant must be ready (getState.isHawkingTenantReady) before the UI state reaches ENABLED.

Critical Constraints

Preconditions: