v6.7.0 proposal

[dd-octo-sts]

Features

• LLM Observability: Add support for OpenAI Agents (trace-processor) #8044

Fixes

• Feature Flags: Allow custom agentless endpoints #9481
• General: Make peer service finalization idempotent #9364
• LLM Observability: Llm observability traces have custom trace IDs #9460
• standalone: Stamp _dd.apm.enabled on every span #9506
• webpack: Opt out of typescript auto resolve #9521

[dd-octo-sts]

Overall package size

Self size: 7.52 MB
Deduped: 8.18 MB
No deduping: 8.18 MB

Dependency sizes

| name | version | self size | total size | |------|---------|-----------|------------| | import-in-the-middle | 3.3.2 | 124.41 kB | 440.65 kB | | opentracing | 0.14.7 | 194.81 kB | 194.81 kB | | dc-polyfill | 0.1.11 | 25.74 kB | 25.74 kB |

_{🤖 This report was automatically generated by heaviest-objects-in-the-universe}

[datadog-datadog-prod-us1-2]

 

⚠️ Warnings

🚦 4 Pipeline jobs failed

All Green | all-green     

Instrumentation | Instrumentation / integration-webpack (node-active)     

Instrumentation | Instrumentation / integration-webpack (node-latest)     

View all 4 failed jobs.

ℹ️ Info

No other issues found (see more)

🧪 All tests passed
❄️ No new flaky tests detected

🎯 Code Coverage (details)
• Patch Coverage: 99.46%
• Overall Coverage: 98.44%

Useful? React with 👍 / 👎

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 60f26cf | Docs | Datadog PR Page | Give us feedback!}

[codecov]

Codecov Report

❌ Patch coverage is 99.45603% with 6 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (v6.x@263c930). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
...ages/datadog-instrumentations/src/openai-agents.js	94.49%	6 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             v6.x    #9518   +/-   ##
=======================================
  Coverage        ?   98.44%           
=======================================
  Files           ?      947           
  Lines           ?   128035           
  Branches        ?    10927           
=======================================
  Hits            ?   126047           
  Misses          ?     1988           
  Partials        ?        0           

Flag	Coverage Δ
aiguard	57.49% <47.72%> (?)
aiguard-integration	56.33% <50.44%> (?)
apm-bucket-0	58.72% <52.27%> (?)
apm-bucket-1	64.02% <52.27%> (?)
apm-bucket-2	62.79% <52.27%> (?)
apm-bucket-3	60.23% <52.27%> (?)
apm-capabilities-tracing	64.30% <34.53%> (?)
apm-integrations-aerospike	56.82% <52.27%> (?)
apm-integrations-confluentinc-kafka-javascript	61.67% <52.27%> (?)
apm-integrations-couchbase	57.18% <47.72%> (?)
apm-integrations-http	62.80% <52.27%> (?)
apm-integrations-kafkajs	62.28% <52.27%> (?)
apm-integrations-next	59.27% <47.72%> (?)
apm-integrations-prisma	58.75% <52.27%> (?)
appsec	72.89% <52.27%> (?)
appsec-express_fastify_graphql	70.41% <52.27%> (?)
appsec-integration	51.50% <45.68%> (?)
appsec-kafka_ldapjs_lodash	64.00% <52.27%> (?)
appsec-mongodb-core_mongoose_mysql	67.72% <52.27%> (?)
appsec-next	57.60% <52.27%> (?)
appsec-node-serialize_passport_postgres	67.39% <52.27%> (?)
appsec-sourcing_stripe_template	65.74% <52.27%> (?)
debugger	64.94% <52.27%> (?)
instrumentations-bucket-0	51.83% <47.72%> (?)
instrumentations-bucket-1	60.32% <47.72%> (?)
instrumentations-bucket-10	62.12% <52.27%> (?)
instrumentations-bucket-11	51.75% <47.72%> (?)
instrumentations-bucket-12	52.31% <47.72%> (?)
instrumentations-bucket-13	51.86% <47.72%> (?)
instrumentations-bucket-2	53.74% <47.72%> (?)
instrumentations-bucket-3	59.32% <47.72%> (?)
instrumentations-bucket-4	52.40% <47.72%> (?)
instrumentations-bucket-5	57.64% <52.27%> (?)
instrumentations-bucket-6	60.85% <52.27%> (?)
instrumentations-bucket-7	58.52% <52.27%> (?)
instrumentations-bucket-8	59.66% <52.27%> (?)
instrumentations-bucket-9	61.64% <47.72%> (?)
instrumentations-instrumentation-couchbase	51.14% <47.72%> (?)
instrumentations-integration-esbuild	34.27% <43.96%> (?)
llmobs-ai_anthropic_bedrock	62.90% <84.61%> (?)
llmobs-bucket-1	61.70% <79.84%> (?)
llmobs-openai	62.44% <48.94%> (?)
llmobs-openai-agents_vertex-ai	59.80% <97.53%> (?)
llmobs-sdk	65.60% <98.82%> (?)
openfeature	56.18% <50.00%> (?)
openfeature-unit	53.48% <52.08%> (?)
platform-core_esbuild_instrumentations-misc	40.64% <45.68%> (?)
platform-integration	61.50% <47.72%> (?)
platform-shimmer_unit-guardrails_webpack	39.13% <45.68%> (?)
plugins-bucket-0	57.15% <52.27%> (?)
plugins-bucket-1	54.48% <47.72%> (?)
plugins-bucket-11	62.36% <47.72%> (?)
plugins-bucket-18	62.06% <52.27%> (?)
plugins-bucket-19	60.15% <47.72%> (?)
plugins-bucket-20	62.15% <47.72%> (?)
plugins-bucket-4	58.68% <47.72%> (?)
plugins-bullmq_cassandra_cookie	61.85% <52.27%> (?)
plugins-cookie-parser_crypto_dd-trace-api	56.83% <47.72%> (?)
plugins-fetch_fs_generic-pool	58.88% <52.27%> (?)
plugins-google-cloud-pubsub_grpc_handlebars	64.78% <52.27%> (?)
plugins-hapi_hono_ioredis	60.32% <52.27%> (?)
plugins-jest_knex_langgraph	55.70% <47.72%> (?)
plugins-ldapjs_light-my-request_limitd-client	58.61% <52.27%> (?)
plugins-lodash_mariadb_memcached	58.19% <52.27%> (?)
plugins-moleculer_mongodb_mongodb-core	62.01% <52.27%> (?)
plugins-mongoose_multer_mysql	59.18% <52.27%> (?)
plugins-mysql2_nats_node-serialize	60.81% <52.27%> (?)
plugins-opensearch_passport-http_pino	59.65% <52.27%> (?)
plugins-postgres_process_pug	58.38% <52.27%> (?)
plugins-redis_router_sequelize	62.20% <52.27%> (?)
plugins-test-and-upstream-rhea_undici_url	61.73% <52.27%> (?)
plugins-valkey_vm_winston	58.12% <52.27%> (?)
plugins-ws	59.70% <47.72%> (?)
profiling	62.24% <52.27%> (?)
serverless-aws-sdk-aws-sdk	55.15% <52.27%> (?)
serverless-aws-sdk-bedrockruntime	54.86% <52.27%> (?)
serverless-aws-sdk-client	56.53% <52.27%> (?)
serverless-aws-sdk-dynamodb	55.76% <52.27%> (?)
serverless-aws-sdk-eventbridge	49.48% <47.72%> (?)
serverless-aws-sdk-kinesis	59.44% <52.27%> (?)
serverless-aws-sdk-lambda	57.49% <52.27%> (?)
serverless-aws-sdk-s3	55.86% <52.27%> (?)
serverless-aws-sdk-serverless-peer-service	59.84% <52.27%> (?)
serverless-aws-sdk-sns	60.30% <52.27%> (?)
serverless-aws-sdk-sqs	60.72% <52.27%> (?)
serverless-aws-sdk-stepfunctions	55.69% <52.27%> (?)
serverless-aws-sdk-util	51.59% <47.72%> (?)
serverless-bucket-0	54.34% <47.72%> (?)
serverless-bucket-1	59.35% <47.72%> (?)
test-optimization-cucumber	71.98% <47.72%> (?)
test-optimization-cypress	65.87% <47.72%> (?)
test-optimization-jest	73.29% <47.72%> (?)
test-optimization-mocha	73.53% <47.72%> (?)
test-optimization-playwright-playwright-atr	60.46% <47.72%> (?)
test-optimization-playwright-playwright-efd	60.65% <47.72%> (?)
test-optimization-playwright-playwright-final-status	60.62% <47.72%> (?)
test-optimization-playwright-playwright-impacted-tests	60.35% <47.72%> (?)
test-optimization-playwright-playwright-reporting	61.64% <47.72%> (?)
test-optimization-playwright-playwright-test-management	61.16% <47.72%> (?)
test-optimization-playwright-playwright-test-span	60.52% <47.72%> (?)
test-optimization-selenium	60.13% <47.72%> (?)
test-optimization-testopt	58.59% <47.72%> (?)
test-optimization-vitest	70.39% <47.72%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[pr-commenter]

Benchmarks

Benchmark execution time: 2026-07-24 06:14:06

Comparing candidate commit 60f26cf in PR branch v6.7.0-proposal with baseline commit 263c930 in branch v6.x.

📊 Benchmarking dashboard

Found 4 performance improvements and 0 performance regressions! Performance is the same for 2310 metrics, 44 unstable metrics.

Explanation

This is an A/B test comparing a candidate commit's performance against that of a baseline commit. Performance changes are noted in the tables below as:

• 🟩 = significantly better candidate vs. baseline
• 🟥 = significantly worse candidate vs. baseline

We compute a confidence interval (CI) over the relative difference of means between metrics from the candidate and baseline commits, considering the baseline as the reference.

If the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD), the change is considered significant.

Feel free to reach out to #apm-benchmarking-platform on Slack if you have any questions.

More details about the CI and significant changes

You can imagine this CI as a range of values that is likely to contain the true difference of means between the candidate and baseline commits.

CIs of the difference of means are often centered around 0%, because often changes are not that big:

---------------------------------(------|---^--------)-------------------------------->
                              -0.6%    0%  0.3%     +1.2%
                                 |          |        |
         lower bound of the CI --'          |        |
sample mean (center of the CI) -------------'        |
         upper bound of the CI ----------------------'

As described above, a change is considered significant if the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD).

For instance, for an execution time metric, this confidence interval indicates a significantly worse performance:

----------------------------------------|---------|---(---------^---------)---------->
                                       0%        1%  1.3%      2.2%      3.1%
                                                  |   |         |         |
       significant impact threshold --------------'   |         |         |
                      lower bound of CI --------------'         |         |
       sample mean (center of the CI) --------------------------'         |
                      upper bound of CI ----------------------------------'

scenario:encoders-0.4-events-native-20

• 🟩 max_rss_usage [-10.947MB; -3.665MB] or [-7.706%; -2.580%]

scenario:llmobs-span-processor-agent-26

• 🟩 cpu_user_time [-514.427ms; -204.853ms] or [-12.777%; -5.088%]
• 🟩 execution_time [-539.037ms; -214.953ms] or [-12.909%; -5.148%]
• 🟩 throughput [+11792.505op/s; +29460.294op/s] or [+5.121%; +12.795%]

Unstable benchmarks

These benchmarks have a confidence interval too wide to call a change; treat them as noise rather than signal.

scenario:appsec-appsec-enabled-24

• unstable execution_time [-209239.094µs; +208885.194µs] or [-7.858%; +7.845%]

scenario:appsec-appsec-enabled-26

• unstable execution_time [-230.049ms; +233.472ms] or [-9.072%; +9.207%]

scenario:appsec-appsec-enabled-with-attacks-24

• unstable execution_time [-162720.199µs; +162277.533µs] or [-5.251%; +5.236%]

scenario:appsec-appsec-enabled-with-attacks-26

• unstable execution_time [-190.531ms; +130.575ms] or [-6.620%; +4.537%]

scenario:appsec-control-20

• unstable execution_time [-115.528ms; +121.379ms] or [-7.107%; +7.467%]

scenario:appsec-control-24

• unstable execution_time [-116270.720µs; +114945.554µs] or [-9.332%; +9.226%]

scenario:appsec-control-26

• unstable execution_time [-118.257ms; +121.989ms] or [-9.603%; +9.907%]

scenario:appsec-iast-no-vulnerability-iast-enabled-always-active-20

• unstable execution_time [-17.540ms; +22.361ms] or [-6.801%; +8.670%]

scenario:appsec-iast-no-vulnerability-iast-enabled-default-config-20

• unstable execution_time [-16273.473µs; +15421.217µs] or [-6.269%; +5.940%]

scenario:debugger-line-probe-with-snapshot-default-24

• unstable cpu_user_time [-1774.221ms; +580.680ms] or [-21.457%; +7.023%]
• unstable execution_time [-1836.006ms; +605.779ms] or [-20.460%; +6.751%]
• unstable instructions [-15.0G instructions; +4.9G instructions] or [-22.170%; +7.186%]
• unstable throughput [-162.034op/s; +482.981op/s] or [-4.405%; +13.129%]

scenario:debugger-line-probe-with-snapshot-default-26

• unstable cpu_user_time [-3674.182ms; +3718.887ms] or [-35.491%; +35.923%]
• unstable execution_time [-3669.094ms; +3727.826ms] or [-33.016%; +33.545%]
• unstable instructions [-32560.9M instructions; +32542.7M instructions] or [-37.673%; +37.652%]
• unstable max_rss_usage [-12131.477KB; +11449.877KB] or [-7.507%; +7.085%]
• unstable throughput [-721.322op/s; +711.449op/s] or [-23.597%; +23.274%]

scenario:debugger-line-probe-with-snapshot-minimal-26

• unstable cpu_user_time [-2459.073ms; +1556.837ms] or [-24.989%; +15.821%]
• unstable execution_time [-2461.507ms; +1537.739ms] or [-23.218%; +14.505%]
• unstable instructions [-21.9G instructions; +13.9G instructions] or [-26.648%; +16.926%]
• unstable throughput [-397.139op/s; +528.406op/s] or [-12.653%; +16.835%]

scenario:debugger-line-probe-without-snapshot-24

• unstable cpu_user_time [-1753.754ms; +564.946ms] or [-21.130%; +6.807%]
• unstable execution_time [-1763.513ms; +566.529ms] or [-19.578%; +6.290%]
• unstable instructions [-15.0G instructions; +4.8G instructions] or [-22.167%; +7.162%]
• unstable throughput [-151.947op/s; +472.095op/s] or [-4.156%; +12.912%]

scenario:dogstatsd-aggregated-20

• unstable cpu_usage_percentage [-8.135%; +5.133%]
• unstable execution_time [-103.215ms; +164.784ms] or [-7.982%; +12.744%]
• unstable throughput [-980046.133op/s; +620377.078op/s] or [-8.337%; +5.277%]

scenario:dogstatsd-aggregated-26

• unstable cpu_usage_percentage [-6.630%; +4.320%]
• unstable execution_time [-63.984ms; +98.660ms] or [-6.306%; +9.724%]
• unstable throughput [-1062690.022op/s; +688173.092op/s] or [-7.093%; +4.593%]

scenario:dogstatsd-with-tags-20

• unstable cpu_user_time [-281.969ms; +429.406ms] or [-5.974%; +9.098%]
• unstable execution_time [-286.908ms; +434.387ms] or [-5.981%; +9.055%]
• unstable throughput [-153407.391op/s; +103679.831op/s] or [-8.762%; +5.922%]

scenario:plugin-graphql-long-with-depth-off-20

• unstable max_rss_usage [-9.640MB; +6.165MB] or [-7.444%; +4.761%]

scenario:plugin-graphql-long-with-depth-off-26

• unstable max_rss_usage [-35.205MB; +24.240MB] or [-16.440%; +11.320%]

scenario:plugin-graphql-long-with-depth-on-max-20

• unstable cpu_user_time [-575.390ms; +604.395ms] or [-4.982%; +5.233%]
• unstable execution_time [-590.130ms; +617.758ms] or [-5.009%; +5.243%]
• unstable throughput [-3.609op/s; +3.436op/s] or [-5.284%; +5.030%]

scenario:plugin-pg-service-26

• unstable cpu_usage_percentage [-8.105%; +6.565%]
• unstable execution_time [-92.354ms; +140.973ms] or [-10.040%; +15.325%]
• unstable throughput [-734723.718op/s; +440307.660op/s] or [-11.000%; +6.592%]

scenario:test-optimization-large-suite-20

• unstable max_rss_usage [-3.319MB; +6.612MB] or [-4.217%; +8.401%]

[datadog-datadog-prod-us1-2]

The new OpenAI Agents LLMObs extraction assumes every Responses array entry is a non-null object. A malformed or partially populated response can therefore throw during annotation and silently lose the LLMObs event, although the surrounding processor prevents the exception from reaching the application.

_{📊 Validated against 12 scenarios · Open Bits AI session}

_{🤖 Datadog Autotest · Commit 60f26cf · What is Autotest? · @DataDog review to ask questions · Any feedback? Reach out in #autotest}

[datadog-datadog-prod-us1-2]

Null Responses items throw during LLMObs extraction

Malformed or forward-compatible Responses payloads lose LLMObs input/output telemetry for the affected model call.

Assertion details

• Input: An OpenAI Agents response or request input contains a null/undefined item in its Responses array, such as output: [null] or input: [null].
• Expected: The instrumentation should ignore malformed array entries or emit a safe placeholder while preserving the APM span and LLMObs event.
• Actual: extractInputMessages and extractOutputMessages dereference item.type without checking item. Direct execution with [null] and [undefined] throws TypeError. DDOpenAIAgentsProcessor catches the exception in onSpanEnd, so the application continues but the LLMObs span is emitted without the intended I/O annotation or may be dropped from processing.

Was this helpful? React 👍 or 👎
_{🤖 Datadog Autotest · What is Autotest? · @DataDog review to ask questions · Any feedback? Reach out in #autotest}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 60f26cf7fc

ℹ️ About Codex in GitHub

Codex has been enabled to automatically 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 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Parent the workflow span to the active trace

When an agent run happens inside an already traced request or job, this startSpan call has no childOf, and DatadogTracer.startSpan() only parents from options.childOf/references. That makes the new Agent workflow span start a separate root trace, so the agent/LLM spans are disconnected from the request trace even though the run happened within it. Use the active legacy/current scope span as the workflow parent when one exists.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Honor OpenAI Agents service overrides

When users configure tracer.use('openai-agents', { service: '...' }) or DD_SERVICE_MAPPING, this plugin's configure() receives config.service, but it only forwards enabled, and these raw tracer.startSpan calls only set component/span.kind. Unlike TracingPlugin.startSpan, nothing writes service.name, so the new workflow/agent/LLM/tool spans are reported under the global service instead of the configured integration service.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Run OpenAI Agents tests on supported Node

In this new job, oldest-maintenance-lts resolves to Node 18 in .github/actions/node/setup/action.yml, but the @openai/agents SDK documents support as Node.js 22 or later (README). Because the openai-agents externals do not declare a node: '>=22' floor, the first install/test leg runs these packages on an unsupported runtime and can fail the LLMObs workflow before the plugin is exercised; run this job on active/latest or add a Node floor to the externals.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Keep the provider with each streamed response

For streamed responses, publishClientBaseURL() runs when the async iterator is created, but the OpenAI Agents response span is not started until the iterator is advanced. If an app creates streams for different base URLs before consuming the first one, this singleton #modelProvider can be overwritten by the later stream, so the earlier LLMObs span is tagged with the wrong provider. Store the provider per model call/span instead of reading the last process-wide value here.

Useful? React with 👍 / 👎.