What problem does it solve? Adding or changing OpenTelemetry telemetry in the Copilot Chat extension requires navigating four different agent execution paths, strict attribute naming conventions, and documentation that must stay in sync with code. This Skill encodes those rules so instrumentation changes are consistent, exportable, and documented. ## Core Features & Use Cases - Span, metric, and event conventions: Enforces GenAI semantic conventions via constants in genAiAttributes.ts, the IOTelService abstraction, and the dual-emit policy for legacy copilot_chat.* attributes. - Agent-specific instrumentation strategies: Covers direct spans for the foreground agent, the bridge SpanProcessor for the in-process Copilot CLI, env var forwarding for terminal CLI, and message-stream synthesis for Claude Code. - Doc synchronization checklists: Requires updating agent_monitoring.md and agent_monitoring_arch.md alongside any code change, plus EXPORTABLE_OPERATION_NAMES for new OTLP-exported operations. - Use Case: When adding a new execute_tool attribute, the Skill directs you to define the constant, truncate values with truncateForOTel, decide on captureContent gating, add tests, and document the attribute in the monitoring docs. ## Quick Start Use the otel skill to add a new span attribute for tool execution latency and update the monitoring docs accordingly.