fix(openai_agents): preserve interrupted workflow spans

[yshapiro-57]

Human Summary

While working on https://github.com/ddoghq/dd-source/pull/120671 I found that the automated evals that are supposed to verify that the new skill is actually working are sporadically producing this error message:

Error preparing LLMObs span event for span <Span(id=3419520971592566224,trace_id=141922899536558176561588771022095746485,parent_id=11792225431829319188,name=Agent workflow)>, missing span kind in span context.

I asked Codex and it came up with a theory that this is because our OpenAI Agents integration is setting kind="workflow" on the agent workflow spans only in on_trace_end and not in on_trace_start. If the code throws an exception before on_trace_end is called, then a span without a kind will be emitted, triggering this error. Codex's suggested fix was to redundantly set kind="workflow" on these spans in on_trace_start.

I am not familiar with the code myself and need to rely on

AI Summary

• Preserve OpenAI Agents workflow spans when a trace is interrupted before normal end-of-trace enrichment.
• Record the workflow classification at trace creation while retaining the existing completion-time input and output enrichment.
• Add regression coverage and a customer-facing release note.

AI Verification

• Full OpenAI Agents integration suite (36 passed, 2 skipped):
  scripts/run-tests --venv 1373e84
• Targeted regression test:
  scripts/run-tests --venv 1373e84 -- tests/contrib/openai_agents/test_openai_agents_llmobs.py -k test_llmobs_workflow_kind_set_on_trace_start
• Repository lint suite:
  scripts/lint checks

Human Verification

Install dd-trace-py from local, rerun the tests from the linked PR in dd-source, and make sure the errors are gone.

Made with Cursor

[cit-pr-commenter-54b7da]

Codeowners resolved as

Resolved from the full PR diff against main using the target branch CODEOWNERS file.
CODEOWNERS team requests not listed below are not required by the current file set.

ddtrace/contrib/internal/openai_agents/processor.py                     @DataDog/ml-observability
ddtrace/llmobs/_integrations/openai_agents.py                           @DataDog/ml-observability
releasenotes/notes/fix-openai-agents-workflow-kind-903518e13a23134c.yaml  @DataDog/apm-python
tests/contrib/openai_agents/test_openai_agents_llmobs.py                @DataDog/ml-observability

[cit-pr-commenter-54b7da]

Circular import analysis

⚠️ Existing circular imports

There are 1 circular imports that already exist on the base branch and have not been changed by this PR.

ddtrace.errortracking._handled_exceptions.bytecode_injector -> ddtrace.errortracking._handled_exceptions.callbacks -> ddtrace.errortracking._handled_exceptions.collector -> ddtrace.errortracking._handled_exceptions.bytecode_reporting -> ddtrace.errortracking._handled_exceptions.bytecode_injector

[cit-pr-commenter-54b7da]

Dependency direction analysis

⚠️ Existing dependency direction violations

There are 201 dependency direction violations that already exist on the base branch and have not been changed by this PR.

Show existing violations (showing 5 of 201 highest severity)

ddtrace.internal.tracemethods -×-> ddtrace.trace  (internal-core -> product:tracing, score=132)
ddtrace.llmobs._integrations.langchain -×-> ddtrace.trace  (product:llmobs -> product:tracing, score=130)
ddtrace.internal.opentelemetry.context -×-> ddtrace.trace  (product:opentelemetry -> product:tracing, score=130)
ddtrace.llmobs._integrations.vllm -×-> ddtrace.trace  (product:llmobs -> product:tracing, score=130)
ddtrace.internal.ci_visibility.filters -×-> ddtrace.trace  (product:ci_visibility -> product:tracing, score=130)

To see all violations, download the layers-base.json and layers-pr.json artifacts from this CI job and run:

uv run --script scripts/import-analysis/layers.py compare layers-base.json layers-pr.json

[datadog-prod-us1-6]

✅ All CI checks and tests passed.

🎉 All green!

🧪 All tests passed
❄️ No new flaky tests detected

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: f3522d6 | Docs | View more details | Give us feedback!}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-07T17:30:21.316751Z	f3522d6	Draft marked ready
🔒 Security Review	✅ Completed 2026-10-07T17:30:46.414390Z	f3522d6	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f3522d6878

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Update the start-time kind guidance

This establishes a new lifecycle requirement—workflow spans must stamp their kind at trace start so interrupted traces can still be exported—but .claude/skills/llmobs-integrations/references/implementation-guide.md still limits its start-time contract to kind="agent" spans. Leaving that reference unchanged means future integrations following the repository guidance can reproduce the same missing-kind drop for workflows; extend the guidance to cover this pattern.

AGENTS.md reference: AGENTS.md:L19-L19

Useful? React with 👍 / 👎.