agentforce-d360-analyze

Trace an Agentforce session in Salesforce Data Cloud by querying materialized DMOs with a 24-query pipeline that resolves sessions by UUID or MessagingSession id and renders a full chronological summary. The skill reconstructs interactions, steps, and gateway requests without manual audit row searches.

Discover Agentforce sessions without an ID by searching with time windows, agent names, channel types, outcomes, or conversation text. The skill resolves matching records across Messaging, Builder, and Voice workflows using timestamp-window fallback logic to locate the relevant session.

Reconstructing Agentforce session behavior requires the Salesforce sf CLI and materialized Data Cloud DMOs. The environment must support the 24-query fetch-assemble-render pipeline to successfully resolve, validate, and trace sessions across Messaging, Builder, and Voice workflows.

Session tracing works with Agentforce Voice and Builder workflows alongside Messaging channels. The pipeline applies resolution and validation logic across all three workflow types, reconstructing generations, LLM gateway requests, feedback, and tags into a unified chronological session summary.

Auditing LLM gateway requests during an Agentforce conversation is best handled by reconstructing the session from Data Cloud artifacts. The skill fetches and assembles gateway requests, generations, and interaction transcripts into an ordered summary for direct review of exact timing and LLM calls.

Agentforce session tracing requires a timestamp-window fallback to resolve sessions when direct UUID or MessagingSession id matching fails. This validation mechanism discovers and reconstructs the correct chronological session behavior by querying materialized Data Cloud DMOs within relevant time ranges.