What problem does it solve? Production Agentforce agents can misroute conversations, fail to invoke actions, or drift from instructions, but these failures are hard to diagnose without visibility into real session data. This Skill queries Session Trace Data Model (STDM) records from Data Cloud, reproduces failures in live preview, and applies targeted fixes to the .agent file. ## Core Features & Use Cases - Session Trace Analysis: Query STDM sessions, conversation details, LLM prompts, aggregated metrics, and RAG quality scores from Data Cloud via a deployable Apex helper class. - Issue Classification: Detect and categorize misroutes, action errors, dead subagents, dead hubs, LOW instruction adherence, knowledge gaps, and safety issues with priority levels. - Reproduce and Fix Workflow: Reproduce confirmed issues with sf agent preview and local traces, then edit the .agent file, validate, publish, and verify with regression test cases. - Use Case: An admin reports that a support agent gives generic answers instead of checking order status. Use this Skill to query the last 7 days of sessions, confirm the action is never invoked, reproduce it in preview, fix the subagent instructions, and publish the corrected agent. ## Quick Start Analyze the recent production sessions of my Agentforce agent in org alias myorg and tell me what is going wrong.
