AI Monitoring
Real-time observability for VS Code Copilot agent sessions, LLM calls, and tool invocations.
Time Range
Users
Team
Model
Tool
Refresh
Usage
Adoption, productivity, delivery, and cost metrics from Copilot agent sessions.
Real-time observability for VS Code Copilot agent sessions, LLM calls, and tool invocations.
Adoption, productivity, delivery, and cost metrics from Copilot agent sessions.
Pull requests created via the CLI agent
Remote agent job PR-ready notifications per second
No data for PR-Ready Rate
No agent activity in this time range. Metrics appear once VS Code Copilot exports OTLP data to Prometheus.
Data source: Copilot cloud PR-ready events
Cloud/remote agent sessions by partner agent
No data for Cloud Session Rate
No cloud agent sessions detected. Remote agent metrics appear once cloud workflows are used.
Data source: Copilot cloud agent sessions
Total new chat sessions in the selected range
Total LLM API calls in the selected range
Rate of new chat sessions per second
Data source: Copilot chat session starts
Lines of code added or removed by accepted agent edits
Data source: Lines added or removed by accepted Copilot edits
Total streaming LLM calls with a first token in the selected range
Input and output token throughput (tokens/s)
Data source: LLM token usage (input and output)
Combined input + output tokens consumed in the selected range
Token throughput per second broken down by LLM model
Data source: LLM token usage by model
Total tool invocations in the selected range
Tool invocations per second by tool name
Data source: Copilot tool invocations
End-to-end agent session duration percentiles (seconds)
Data source: Copilot agent session duration
LLM round-trips per agent invocation (p50/p90/p99)
Data source: Copilot agent LLM turns per session
4-gram similarity score of AI code that persists (0–1)
Data source: Copilot code survival (4-gram similarity)
Tool execution latency percentiles (milliseconds)
Data source: Copilot tool execution time