Amazon CloudWatch Launches Coding Agent Insights for AI Coding Tools
Amazon CloudWatch's new Coding Agent Insights gives engineering leaders telemetry-based visibility into AI coding tool usage.
What it does
Amazon CloudWatch rolled out Coding Agent Insights to give engineering leaders visibility into how their AI coding tools are actually driving value across the organization. The feature connects to the Claude apps gateway for AWS, pulling telemetry straight from Claude Code without requiring any extra instrumentation on a team's part. Codex and GitHub Copilot are the other two coding agents CloudWatch currently supports.
Why AWS built it
AWS frames this as a return-on-investment problem: organizations scaling their AI coding agent adoption need a way to measure what they are getting back. The feature runs on OpenTelemetry metrics that an organization's coding agents already emit, layering those metrics onto the CloudWatch operational data that organization already has, rather than presenting a separate, disconnected dashboard.
What questions it answers
AWS says the feature helps engineering leaders answer practical questions such as which teams would gain the most from wider access, where agents are speeding up delivery, and how to size token budgets sensibly across departments. From there, teams can watch spend trends over time, set proactive token billing alerts, link agent use to gains in commit throughput and pull-request speed, and find which models give the best cost-to-output ratio for their workloads.
Availability and setup
CloudWatch Coding Agent Insights is available in all AWS commercial regions, with three named exceptions: Middle East (UAE), Middle East (Bahrain), and Israel (Tel Aviv). To turn it on, an organization points its Claude apps gateway at CloudWatch to start emitting telemetry, following AWS's setup guide, then opens the Coding Agent Insights dashboard directly in the CloudWatch console. AWS says standard CloudWatch OpenTelemetry metric ingestion pricing applies, and directs customers to its separate metrics pricing page for the specifics. AWS also points readers to its documentation for anyone who wants to learn more about the feature beyond the setup guide and the pricing page.
What it does not cover
The announcement is specific about scope: Coding Agent Insights is a monitoring and reporting layer, not a way to control which coding agent a developer uses day to day. It reads OpenTelemetry metrics that Claude Code, Codex, and GitHub Copilot already emit; it does not claim to change how those agents behave. For teams already emitting OpenTelemetry data, AWS frames the added work as configuration, not new instrumentation, since the Claude apps gateway for AWS handles the Claude Code telemetry path directly.
Who this is for
Coding Agent Insights is aimed at engineering leaders and platform teams overseeing AI coding agent rollouts, not individual developers using Claude Code, Codex, or GitHub Copilot day to day. The dashboard lives inside CloudWatch, the same console AWS customers already use for operational monitoring, so a leader does not need to learn a new tool to see coding-agent telemetry alongside everything else they track.
Key facts
- Coding Agent Insights integrates with the Claude apps gateway for AWS to collect Claude Code telemetry without extra instrumentation.
- Codex and GitHub Copilot are also supported coding agents.
- The feature is built on OpenTelemetry metrics, shown alongside existing CloudWatch operational data.
- It is available in all AWS commercial regions except Middle East (UAE), Middle East (Bahrain), and Israel (Tel Aviv).
- Standard CloudWatch OpenTelemetry metric ingestion pricing applies, per AWS's metrics pricing page.
Check your understanding
Which three coding agents does AWS name as supported by Coding Agent Insights? Which regions are excluded from CloudWatch Coding Agent Insights availability?