Update OpenTelemetry GenAI conventions to latest

[github-actions]

Draft pending the first release of
open-telemetry/semantic-conventions-genai.
Both conventions are Development stability and unreleased. Merge once they ship in a tagged release.

What this PR implements

Area	Convention	Upstream	Compensating change
gen-ai	document modality in multimodal content parts	#142	Added IsDocumentMediaType in OtelMessageSerializer; DeriveModalityFromMediaType returns "document" for PDF/Office/ODF MIME types. Augmented AllOfficialOtelContentPartTypes_SerializedCorrectly test.
gen-ai	gen_ai.request.reasoning.level attribute	#258	Added ReasoningLevel constant in OpenTelemetryConsts and emission in OpenTelemetryChatClient; maps ChatOptions.Reasoning?.Effort to "low"/"medium"/"high"/"extra_high". Augmented ExpectedInformationLogged_Async.
gen-ai	Doc-comment wording migration	—	Updated all OpenTelemetry* client files from Semantic Conventions for Generative AI systems v1.41 to GenAI Semantic Conventions latest.

Validation: build clean (net8.0/net9.0/net10.0, 0 warnings); 15 OpenTelemetryChatClientTests pass (all 3 TFMs). No public API surface change (OpenTelemetryConsts is internal); doc-comment at GenAI Semantic Conventions latest.

Upstream scan tracking

Legend: 🔴 implemented here · ✅ already aligned · 🟡 watch/deferred · 🟢 not applicable (no client / docs-only)

Merged upstream changes (Unreleased) -- applicability to dotnet/extensions

Scanned at commit 93de98a (2026-06-18).

Upstream PR	Area	Change	Applicability	Status
#97	gen-ai	plan operation name for agent planning spans	🟡	Deferred — no planning span today.
#99	gen-ai	moonshot_ai well-known value for gen_ai.provider.name	🟢	No code change needed.
#107	gen-ai	gen_ai.agent.name marked sampling-relevant	🟢	No emission site in this repo.
#126	gen-ai	gen_ai.workflow.duration metric + gen_ai.workflow.name	🟢	No workflow instrumentation today.
#136	mcp	gen_ai.tool.call.arguments / result on MCP server span	🟢	No MCP server instrumentation.
#140	gen-ai	GenAI memory operations + attributes	🟢	No memory client instrumentation today.
#142	gen-ai	document modality in multimodal content parts	🔴	Implemented in OtelMessageSerializer.DeriveModalityFromMediaType.
#162	gen-ai	gen_ai.conversation.compacted + CompactionPart schema	🟡	Deferred — no compaction support today.
#211	gen-ai	Clarify billed vs. consumed token counts	🟢	Docs-only clarification.
#212	gen-ai	Generalize gen_ai.provider.name description	🟢	Docs-only clarification.
#214	gen-ai	gen_ai.provider.name relaxed to Conditionally Required	✅	Already aligned — emitted when available.
#216	gen-ai	Span SHOULD cover full duration including retries	🟢	Docs-only clarification.
#217	gen-ai	gen_ai.request.top_k type double → int, decoding-only	✅	Already aligned — uses if (options.TopK is int topK).
#217	gen-ai	Clarify top_k is sampling/decoding, not top_logprobs	🟢	Docs-only clarification.
#219	gen-ai	gen_ai.conversation.id must not use fallback values	🟢	Docs-only clarification.
#220	mcp	Clarify MCP context propagation per SEP-414	🟢	No MCP instrumentation.
#242	gen-ai	gen_ai.agent.id scope clarified; removed from internal spans	✅	Already aligned — not emitted in this repo.
#257	gen-ai	Limit gen_ai.system_instructions parts to text only	✅	Already aligned — serialized as text-only.
#258	gen-ai	gen_ai.request.reasoning.level added to inference spans	🔴	Implemented in OpenTelemetryChatClient; maps ChatOptions.Reasoning?.Effort.
#289	gen-ai	Remove gen_ai.provider.name from invoke_agent internal span	🟢	No invoke_agent instrumentation today.

In-flight upstream changes (open PRs) -- applicability if merged

Excluded dependency / CI / chore / server-only PRs: dotnet#112, dotnet#164, dotnet#202, dotnet#215, dotnet#252, dotnet#283, dotnet#290, dotnet#291, dotnet#324, dotnet#325, dotnet#328, dotnet#340, dotnet#351.

Upstream PR	Area	Change	Applicability	Status
#96	gen-ai	gen_ai.token.cache / gen_ai.token.reasoning metric attributes	🟡	Watch — new breakdown attributes on gen_ai.client.token.usage.
#98	gen-ai	Agent-to-agent handoff as execute_tool span	🟡	Watch — may affect FunctionInvocationProcessor span naming.
#143	gen-ai	Optional byte_size on multimodal content parts	🟡	Watch — additive field on OtelBlobPart / OtelUriPart / OtelFilePart.
#144	gen-ai	BlobPart.content optional + stripped_reason	🟡	Watch — may require OtelBlobPart.Content to be made optional.
#197	gen-ai	Modality / cache / phase token-usage breakdown metrics	🟡	Watch — additive token-usage breakdown.
#238	gen-ai	gen_ai.agent.finish_reason attribute	🟡	Watch — no invoke_agent instrumentation today.
#250	gen-ai	gen_ai.agent.invocation.id attribute	🟡	Watch — no invoke_agent instrumentation today.
#270	gen-ai	GenAI agent entity attributes	🟡	Watch — relevant when agent spans land.
#336	gen-ai	gen_ai.invoke_agent.inference_calls / tool_calls metrics	🟡	Watch — no invoke_agent instrumentation today.
#341	gen-ai	Rename gen_ai.workflow.duration → gen_ai.invoke_workflow.duration	🟡	Watch — metric name change if workflow instrumentation lands.
#350	gen-ai	gen_ai.agent.identity.id / gen_ai.agent.identity.name	🟡	Watch — relevant when agent spans land.

Tracking state

# otel-genai-tracking:begin
Upstream-Repo: open-telemetry/semantic-conventions-genai
Upstream-Scan-Ref: 93de98a5b1298ee6708c76ca87cb389861689732  # 2026-06-18; gen_ai.request.reasoning.level (#258) integrated
Upstream-Scan-Date: 2026-06-28T01:39:36Z
Upstream-Release: none                        # Unreleased; Towncrier fragments under changelog.d/
Core-Semconv-Dependency: v1.42.0              # versions.env SEMCONV_VERSION (core dep, NOT the GenAI version)
DotnetExtensions-Implemented-Version: latest  # doc-comment migrated to GenAI Semantic Conventions latest
# otel-genai-tracking:end

Generated by MEAI: Update Otel Gen-AI · 224.7 AIC · ⊞ 10.8K · ◷

[github-actions]

Commit pushed: 83fe9f1

Generated by MEAI: Update Otel Gen-AI · 224.7 AIC · ⊞ 10.8K

[github-actions]

Incremental update: integrated upstream semantic-conventions-genai#258 — scanned commit 93de98a (2026-06-18).

New convention implemented:

• gen_ai.request.reasoning.level — added ReasoningLevel constant to OpenTelemetryConsts.GenAI.Request and emission in OpenTelemetryChatClient; maps ChatOptions.Reasoning?.Effort (Low/Medium/High/ExtraHigh) to the strings "low"/"medium"/"high"/"extra_high" respectively. ReasoningEffort.None does not emit the tag.

Validation: 15 OpenTelemetryChatClientTests pass on net8.0/net9.0/net10.0 (0 build warnings). No public API surface change.

Generated by MEAI: Update Otel Gen-AI · 224.7 AIC · ⊞ 10.8K · ◷