Compatibility Reference

Use this document for version-sensitive guidance that changes more frequently than the core routing logic in SKILL.md.

Baseline version floors

• OpenTelemetry Collector: v0.153.0+ (the first release that supports every canonical component ID used by this skill, including load_balancing)
• Core Semantic Conventions: v1.40.0+
• GenAI Semantic Conventions: follow the separate open-telemetry/semantic-conventions-genai repository; the signal definitions are Development and do not currently have a stable release floor
• Kubernetes: native restartable sidecars are enabled by default from v1.29; ordinary multi-container Pods are a separate pattern
• Go SDK: v1.24.0+
• Python SDK: v1.41.0+ for unchanged examples; the new event/GenAI examples target v1.44.0 (instrumentation release train 0.65b0). See Python guidance.
• Collector 0.160 examples: selective Prometheus resource_constant_labels requires v0.160.0; custom builds require Go 1.26. The general canonical-ID floor remains v0.153.0.
• Python GenAI packages: migrated to open-telemetry/opentelemetry-python-genai; released beta versions and supported libraries are listed in the Python reference. Do not substitute unreleased skeletons.

AI agent telemetry compatibility

• Claude Code: current release emits metrics plus logs/events and beta traces with selected GenAI attributes (for example gen_ai.tool.call.id), not full schema alignment; OTEL_METRICS_INCLUDE_ENTRYPOINT=true adds optional bounded app.entrypoint
• Gemini CLI: v0.34.0+ emits traces, metrics, and logs with GenAI semantic conventions
• GitHub Copilot: latest stable / Insiders builds expose traces, metrics, and events with GenAI semantic conventions
• Codex CLI: current documentation describes structured log events and metrics for API requests, tool calls, exec, and sessions; verify trace support and mode-specific behavior in the installed release
• Qwen Code: v0.16.1+ emits traces, metrics, and logs with partial gen_ai.* dual-emit layered on top of authoritative qwen-code.* fields

Maintenance guidance

• Treat these version floors as fast-moving compatibility notes rather than hard-coded architectural rules.
• Pin GenAI dashboards and transforms to the fields actually emitted by the agent/instrumentation version; do not infer current GenAI behavior from the core Semantic Conventions version alone.
• Pin collector components to released versions and verify stability levels before using non-stable features in production.
• Re-check upstream release notes whenever updating examples that depend on AI agent telemetry support or evolving semantic conventions.