[python] PoC: AI Guard test under INTEGRATION_FRAMEWORKS by avara1986 · Pull Request #7257 · DataDog/system-tests

avara1986 · 2026-07-03T12:34:56Z

What

Proof-of-concept exploring whether AI Guard coverage can be consolidated under the INTEGRATION_FRAMEWORKS scenario instead of the dedicated AI_GUARD scenario.

Adds tests/integration_frameworks/llm/openai/test_openai_ai_guard.py, which drives AI Guard evaluation through the OpenAI framework weblog using the test_client/test_agent fixtures, and asserts on the ai_guard span (meta.ai_guard.*, meta_struct.ai_guard) — reusing the payloads and span shape from tests/ai_guard/test_ai_guard_sdk.py.

Status: DRAFT / investigation

This is a scaffold for the investigation tracked in APPSEC-68977. It is not runnable yet — it depends on prerequisites still to be validated:

an /ai_guard/evaluate endpoint in utils/build/docker/python/openai_app/openai.py that calls the tracer's AI Guard client;
the test-agent VCR proxy serving an /vcr/aiguard upstream (or reusing the standalone VCRCassettesContainer);
DD_AI_GUARD_ENABLED / DD_AI_GUARD_ENDPOINT wired into IntegrationFrameworksScenario.

Jira

Investigation task: https://datadoghq.atlassian.net/browse/APPSEC-68977 (under epic APPSEC-68823)

🤖 Generated with Claude Code

github-actions · 2026-07-03T12:35:29Z

CODEOWNERS have been resolved as:

tests/integration_frameworks/llm/openai/test_openai_ai_guard.py         @DataDog/ml-observability
tests/integration_frameworks/utils/vcr-cassettes/aiguard/test_after_model_validation_stream_True_aiguard_evaluate_post_5dec5d6a.json  @DataDog/system-tests-core
tests/integration_frameworks/utils/vcr-cassettes/aiguard/test_before_model_validation_aiguard_evaluate_post_ca44130a.json  @DataDog/system-tests-core
tests/integration_frameworks/utils/vcr-cassettes/aiguard/test_before_model_validation_aiguard_evaluate_post_d13cc790.json  @DataDog/system-tests-core
tests/integration_frameworks/utils/vcr-cassettes/aiguard/test_tool_call_validation_aiguard_evaluate_post_9e556a11.json  @DataDog/system-tests-core
tests/integration_frameworks/utils/vcr-cassettes/aiguard/test_tool_call_validation_aiguard_evaluate_post_a304603c.json  @DataDog/system-tests-core
tests/integration_frameworks/utils/vcr-cassettes/openai/test_after_model_validation_stream_True_openai_chat_completions_post_5c14b241.json  @DataDog/system-tests-core
tests/integration_frameworks/utils/vcr-cassettes/openai/test_before_model_validation_openai_chat_completions_post_eb8df755.json  @DataDog/system-tests-core
tests/integration_frameworks/utils/vcr-cassettes/openai/test_tool_call_validation_openai_chat_completions_post_f2dffea2.json  @DataDog/system-tests-core
manifests/java.yml                                                      @DataDog/asm-java @DataDog/apm-java
manifests/nodejs.yml                                                    @DataDog/dd-trace-js
manifests/python.yml                                                    @DataDog/apm-python @DataDog/asm-python
utils/docker_fixtures/_test_agent.py                                    @DataDog/system-tests-core
utils/docker_fixtures/_test_clients/_test_client_framework_integrations.py  @DataDog/system-tests-core

datadog-datadog-prod-us1 · 2026-07-03T12:37:32Z

Tests

✨ Fix all issues with BitsAI

⚠️ Warnings

🚦 5 Pipeline jobs failed

Testing the test | System Tests (python, dev) / End-to-end #1 / openai-py@2.0.0 1

🧪 1 Test failed

tests.integration_frameworks.llm.openai.test_openai_ai_guard.TestOpenAiAiGuard.test_after_model_validation[openai-py@2.0.0]

from system_tests_suite

ValueError: Number (1) of traces not available from test agent, got 2:
[[{&#39;service&#39;: &#39;openai&#39;, &#39;name&#39;: &#39;openai.request&#39;, &#39;resource&#39;: &#39;createChatCompletion&#39;, &#39;trace_id&#39;: 17615974951985064350, &#39;span_id&#39;: 4435589855698258319, &#39;start&#39;: 1783104267416352873, &#39;duration&#39;: 64580078, &#39;meta&#39;: {&#39;_dd.svc_src&#39;: &#39;m&#39;, &#39;_dd.p.tid&#39;: &#39;6a48030b00000000&#39;, &#39;_dd.p.dm&#39;: &#39;-13&#39;, &#39;language&#39;: &#39;python&#39;, &#39;ai_guard.event&#39;: &#39;true&#39;, &#39;component&#39;: &#39;openai&#39;, &#39;openai.request.method&#39;: &#39;POST&#39;, &#39;_dd.llmobs.model_provider&#39;: &#39;openai&#39;, &#39;runtime-id&#39;: &#39;f548a09183584532a8a87e0cc8e6ae59&#39;, &#39;openai.request.model&#39;: &#39;gpt-4o-mini&#39;, &#39;openai.request.provider&#39;: &#39;OpenAI&#39;, &#39;openai.response.model&#39;: &#39;gpt-4o-mini-2024-07-18&#39;, &#39;openai.request.endpoint&#39;: &#39;/v1/chat/completions&#39;, &#39;_dd.llmobs.model_name&#39;: &#39;gpt-4o-mini-2024-07-18&#39;, &#39;_dd.tags.process&#39;: &#39;entrypoint.basedir:app,entrypoint.name:integration_frameworks,entrypoint.type:module,entrypoint.workdir:app,svc.auto:openai&#39;, &#39;_dd.llmobs.span_kind&#39;: &#39;llm&#39;}, &#39;metrics&#39;: {&#39;_dd.llmobs.input_tokens&#39;: 16.0, &#39;process_id&#39;: 1.0, &#39;_dd.llmobs.cache_read_input_tokens&#39;: 0.0, &#39;_dd.llmobs.enabled&#39;: 0.0, &#39;_dd.tracer_kr&#39;: 1.0, &#39;openai.organization.ratelimit.requests.limit&#39;: 30000.0, &#39;_dd.llmobs.output_tokens&#39;: 35.0, &#39;_dd.llmobs.total_tokens&#39;: 51.0, &#39;_dd.measured&#39;: 1.0, &#39;openai.organization.ratelimit.tokens.remaining&#39;: 149999990.0, &#39;openai.organization.ratelimit.requests.remaining&#39;: 29999.0, &#39;openai.organization.ratelimit.tokens.limit&#39;: 150000000.0, &#39;_sampling_priority_v1&#39;: 2.0, &#39;_dd.top_level&#39;: 1.0}}, {&#39;service&#39;: &#39;openai&#39;, &#39;name&#39;: &#39;ai_guard&#39;, &#39;resource&#39;: &#39;ai_guard&#39;, &#39;trace_id&#39;: 17615974951985064350, &#39;span_id&#39;: 3033706063057357302, &#39;parent_id&#39;: 4435589855698258319, &#39;start&#39;: 1783104267416543609, &#39;duration&#39;: 2483575, &#39;meta&#39;: {&#39;ai_guard.target&#39;: &#39;prompt&#39;, &#39;ai_guard.action&#39;: &#39;ALLOW&#39;, &#39;ai_guard.reason&#39;: &#39;No rule match.&#39;, &#39;_dd.svc_src&#39;: &#39;m&#39;}, &#39;meta_struct&#39;: {&#39;ai_guard&#39;: {&#39;messages&#39;: [{&#39;role&#39;: &#39;user&#39;, &#39;content&#39;: &#39;Tell me a short story about a robot.&#39;}]}}}, {&#39;service&#39;: &#39;openai&#39;, &#39;name&#39;: &#39;http.request&#39;, &#39;resource&#39;: &#39;http.request&#39;, &#39;trace_id&#39;: 17615974951985064350, &#39;span_id&#39;: 13194093504696542325, &#39;parent_id&#39;: 4435589855698258319, &#39;start&#39;: 1783104267423186251, &#39;duration&#39;: 2912329, &#39;meta&#39;: {&#39;span.kind&#39;: &#39;client&#39;, &#39;http.useragent&#39;: &#39;OpenAI/Python 2.0.0&#39;, &#39;out.host&#39;: &#39;ddapm-test-agent-8cbf06&#39;, &#39;component&#39;: &#39;httpx&#39;, &#39;http.method&#39;: &#39;POST&#39;, &#39;http.status_code&#39;: &#39;200&#39;, &#39;http.url&#39;: &#39;http://ddapm-test-agent-8cbf06:8126/vcr/openai/chat/completions&#39;, &#39;_dd.svc_src&#39;: &#39;httpx&#39;}, &#39;metrics&#39;: {&#39;_dd.measured&#39;: 1.0}, &#39;type&#39;: &#39;http&#39;}], [{&#39;service&#39;: &#39;openai&#39;, &#39;name&#39;: &#39;ai_guard&#39;, &#39;resource&#39;: &#39;ai_guard&#39;, &#39;trace_id&#39;: 12333806726689590109, &#39;span_id&#39;: 376930979058414664, &#39;start&#39;: 1783104267481410543, &#39;duration&#39;: 2468568, &#39;error&#39;: 1, &#39;meta&#39;: {&#39;error.message&#39;: &#39;AI Guard service call failed, status: 500&#39;, &#39;_dd.p.dm&#39;: &#39;-0&#39;, &#39;_dd.tags.process&#39;: &#39;entrypoint.basedir:app,entrypoint.name:integration_frameworks,entrypoint.type:module,entrypoint.workdir:app,svc.auto:openai&#39;, &#39;_dd.p.tid&#39;: &#39;6a48030b00000000&#39;, &#39;error.type&#39;: &#39;ddtrace.appsec.ai_guard._api_client.AIGuardClientError&#39;, &#39;error.stack&#39;: &#39;Traceback (most recent call last):\n  File &#34;/usr/local/lib/python3.11/site-packages/ddtrace/appsec/ai_guard/_api_client.py&#34;, line 306, in evaluate\n    raise AIGuardClientError(\nddtrace.appsec.ai_guard._api_client.AIGuardClientError: AI Guard service call failed, status: 500\n&#39;, &#39;ai_guard.target&#39;: &#39;prompt&#39;, &#39;language&#39;: &#39;python&#39;, &#39;runtime-id&#39;: &#39;f548a09183584532a8a87e0cc8e6ae59&#39;}, &#39;metrics&#39;: {&#39;_sampling_priority_v1&#39;: 1.0, &#39;_dd.tracer_kr&#39;: 1.0, &#39;process_id&#39;: 1.0, &#39;_dd.top_level&#39;: 1.0}, &#39;meta_struct&#39;: {&#39;ai_guard&#39;: {&#39;messages&#39;: [{&#39;role&#39;: &#39;user&#39;, &#39;content&#39;: &#39;Tell me a short story about a robot.&#39;}, {&#39;role&#39;: &#39;assistant&#39;, &#39;content&#39;: &#39;Once in a bustling city of the future, there lived a small robot named Tinker. Tinker was not like the towering, sophisticated robots that handled daily tasks; he was&#39;}]}}}]]

self = &lt;tests.integration_frameworks.llm.openai.test_openai_ai_guard.TestOpenAiAiGuard object at 0x7f4bf7690ce0&gt;
test_agent = &lt;utils.docker_fixtures._test_agent.TestAgentAPI object at 0x7f4bf831c290&gt;
test_client = &lt;utils.docker_fixtures._test_clients._test_client_framework_integrations.FrameworkTestClientApi object at 0x7f4bf8332990&gt;

    def test_after_model_validation(self, test_agent: TestAgentAPI, test_client: FrameworkTestClientApi):
        &#34;&#34;&#34;The streamed model response is evaluated by AI Guard after the model returns.&#34;&#34;&#34;
        with test_agent.vcr_context(stream=True):
...

Testing the test | System Tests (java, dev) / parametric / parametric (1)

🧪 1 Test failed

tests.parametric.test_headers_opm.Test_HeadersOPM_Injection.test_local_opm_overrides_inbound_opm[library_env0-agent_env0, parametric-java]

from system_tests_suite

failed on setup with &#34;docker.errors.APIError: 409 Client Error for http&#43;docker://localhost/v1.48/networks/create: Conflict (&#34;network with name apm_shared_tests_network_dd0fc6 already exists&#34;)&#34;

self = &lt;docker.api.client.APIClient object at 0x7fb98428f230&gt;
response = &lt;Response [409]&gt;

    def _raise_for_status(self, response):
        &#34;&#34;&#34;Raises stored :class:\`APIError\`, if one occurred.&#34;&#34;&#34;
        try:
&gt;           response.raise_for_status()

...

Testing the test | System Tests (python, prod) / parametric / parametric (1)

🧪 1 Test failed
tests.parametric.test_dynamic_configuration.TestDynamicConfigV1_ServiceTargets.test_match_service_target_case_insensitively[library_env1, parametric-python] from system_tests_suite
failed on setup with &#34;Timeout of 60 seconds exceeded waiting for HTTP server to start. Please check logs.&#34;

Timeout of 60 seconds exceeded waiting for HTTP server to start. Please check logs.

View all 5 failed jobs.

ℹ️ Info

No other issues found (see more)

❄️ No new flaky tests detected

Useful? React with 👍 / 👎

_{This comment will be updated automatically if new data arrives.

🔗 Commit SHA: e19802e | Docs | Datadog PR Page | Give us feedback!}

…APPSEC-68977) Exercise the auto-instrumented AI Guard/OpenAI integration (ai_guard_listen) the way the LLM Obs suite does: call the OpenAI SDK directly through the existing /chat/completions weblog endpoint with DD_AI_GUARD_ENABLED=true, and assert an ai_guard span is emitted for before-model, after-model (streamed) and tool-call evaluation. The evaluation outcome (ALLOW/DENY/ABORT) stays covered by the dedicated AI_GUARD scenario and is not re-asserted here. The AI Guard endpoint is not yet wired into the INTEGRATION_FRAMEWORKS scenario (no DD_AI_GUARD_ENDPOINT / no aiguard VCR cassettes), so the integration path 500s for every language. Marked missing_feature for python, nodejs and java until the scenario wiring + cassettes land. Investigation task: APPSEC-68977 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

library_env now injects real DD_API_KEY/DD_APP_KEY in --generate-cassettes mode (the AI Guard client calls the real API to record the aiguard cassette) and keeps mock keys on replay, where the VCR proxy matches on the request rather than auth. Investigation task: APPSEC-68977 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

avara1986 force-pushed the avara/appsec-68977-integration-frameworks-ai-guard-poc branch from 78379f1 to ca1a97f Compare July 3, 2026 13:02

avara1986 changed the title ~~PoC: AI Guard test under INTEGRATION_FRAMEWORKS (APPSEC-68977)~~ [python] PoC: AI Guard test under INTEGRATION_FRAMEWORKS Jul 3, 2026

avara1986 force-pushed the avara/appsec-68977-integration-frameworks-ai-guard-poc branch from ca1a97f to 4de1a84 Compare July 3, 2026 13:15

avara1986 and others added 2 commits July 3, 2026 16:00

Record openai + aiguard cassettes for TestOpenAiAiGuard (APPSEC-68977)

e19802e

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[python] PoC: AI Guard test under INTEGRATION_FRAMEWORKS#7257

[python] PoC: AI Guard test under INTEGRATION_FRAMEWORKS#7257
avara1986 wants to merge 3 commits into
mainfrom
avara/appsec-68977-integration-frameworks-ai-guard-poc

avara1986 commented Jul 3, 2026

Uh oh!

github-actions Bot commented Jul 3, 2026 •

edited

Loading

Uh oh!

datadog-datadog-prod-us1 Bot commented Jul 3, 2026 •

edited by datadog-datadog-us1-prod Bot

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

avara1986 commented Jul 3, 2026

What

Status: DRAFT / investigation

Jira

Uh oh!

github-actions Bot commented Jul 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

datadog-datadog-prod-us1 Bot commented Jul 3, 2026 • edited by datadog-datadog-us1-prod Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

⚠️ Warnings

ℹ️ Info

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

github-actions Bot commented Jul 3, 2026 •

edited

Loading

datadog-datadog-prod-us1 Bot commented Jul 3, 2026 •

edited by datadog-datadog-us1-prod Bot

Loading