Add Cloudflare Agents guide to LLM Observability OpenTelemetry instrumentation

[Kyle-Verhoog]

What does this PR do? What is the motivation?

Adds a "Using Cloudflare Agents" guide to the LLM Observability OpenTelemetry instrumentation page. Cloudflare Agents on Workers emit GenAI-convention spans, but the export path differs from every other framework on the page: the OTLP endpoint and headers are set on a destination in the Cloudflare dashboard rather than with OTEL_EXPORTER_OTLP_* environment variables. The guide covers enabling tracing in the Worker config, creating the Datadog destination with the dd-otlp-source=llmobs header, and wrapping the Vercel AI SDK with wrapAISDK() so model calls produce chat/invoke_agent spans. Also adds Cloudflare Agents to the Node.js tested-frameworks table.

Validated end-to-end: traces from a Cloudflare Agent land in Agent Observability with model, token usage, and input/output messages populated.

Merge readiness

• Ready for merge

AI assistance

Drafted with Claude Code after validating the integration end-to-end against a live Cloudflare Agent and Datadog Agent Observability.

[github-actions]

Preview links (active after the build_preview check completes)

Modified Files

• https://docs-staging.datadoghq.com/kyle.verhoog/docs-llmobs-otel-cloudflare/llm_observability/instrument/otel_instrumentation

[janine-c]

Hey Kyle, looks great! I made some minor formatting changes so the steps you wrote would render as a numbered list, which I think makes it more readable, and a couple other minor formatting changes. I'll go ahead and merge for you when the checks pass 🙂 Thanks for adding this!