Live Feed/Sentry/Fact Record
Sentry logo
Sentry
product launch 96% Confidence Gate September 11, 2026

Agent Tracing is now GA

Sentry has moved Agent Tracing from beta to General Availability. The feature now provides visibility into individual model calls, tool executions, and agent handoffs within existing trace telemetry.

Verified State Diff

Comparison Mode:
- Previous State
Agentic workflows were treated as black boxes, lacking granular visibility into internal model calls and tool execution steps within Sentry traces.
+ Verified New State
Full observability into model calls, tool executions, and agent handoffs is now natively supported and generally available for production use.

Impact & Verification Analysis

WHO IS AFFECTED

Software engineers and AI developers building applications using LLM agents and tool-calling frameworks.

WHY IT MATTERS

It enables production-grade debugging for AI agents, allowing developers to identify latency bottlenecks or logic failures in complex, multi-step agentic interactions.

Full Fact Overview

This release integrates AI-specific observability directly into Sentry's distributed tracing infrastructure. By capturing the full user-agent exchange, developers can now debug non-deterministic agentic workflows, monitor token usage, and inspect tool-calling sequences within the 'Conversations' interface. This bridges the gap between traditional application performance monitoring (APM) and the opaque nature of LLM-based agent execution chains.

Multi-Source Evidence Chain (1)

Agent Tracing is now GASentry
TRACKED ENTITY
Explore all historical Sentry changes
View Sentry Hub ➔