Extracts and analyzes Agentforce session telemetry to debug agent conversations and optimize performance using Data Cloud STDM data.
The sf-ai-agentforce-observability skill provides deep visibility into Salesforce Agentforce by extracting Session Tracing Data Model (STDM) records and analyzing them using high-performance tools like Polars and Parquet. It enables developers to reconstruct multi-turn session timelines, identify routing failures, and pinpoint latency hotspots within agentic workflows. By leveraging Data 360 telemetry, this skill moves beyond basic testing into comprehensive, data-driven observability, allowing for precise root-cause analysis of topic mismatches and action failures.
主な機能
01Automated STDM data extraction from Data 360 APIs
02Reconstruction of multi-turn session timelines and interaction steps
03Integration with GenAI Trust Layer for content quality auditing
04Latency and error identification for topic routing and agent actions
05198 GitHub stars
06High-speed telemetry analysis using Polars and Parquet storage
ユースケース
01Auditing Agentforce conversation history for compliance and generation quality
02Debugging why an agent failed to route a specific user query to the correct topic
03Analyzing performance bottlenecks and latency hotspots in complex agent interactions
