[Chad Palmer (chadpalmer)]

Summary

This PR includes three related updates for the 0.261.061 branch:

• Excel / tabular processing fixes

  • Adds opt-in HTTP attempt diagnostics for foreground tabular Azure OpenAI calls, including retry counts, timings, correlation IDs, and failure metadata.
  • Preserves required-tool behavior across tabular analysis rounds so the model can complete after initial tool usage while still allowing dependent follow-up tool rounds.
  • Improves tabular retry test isolation so helper tests no longer reach live Cosmos configuration during local validation.
  • Ref: functions_tabular_model_diagnostics.py, functions_model_endpoint_runtime.py, route_backend_chats.py, test_tabular_model_http_diagnostics.py, test_tabular_tool_error_retry_and_thoughts.py

• Content safety admin settings update

  • Adds configurable content safety category thresholds in admin settings.
  • Updates admin content safety UI and documentation for the new settings.
  • Enhances safety violation admin list/card views and CSV exports with date and user metadata.
  • Ref: functions_content_safety.py, functions_settings.py, route_backend_safety.py, admin_settings.js, admin-safety-violations.js, CONTENT_SAFETY_BLOCK_THRESHOLD.md, SAFETY_VIOLATION_METADATA_DISPLAY_FIX.md

• Workflow instructions draft button fix

  • Updates workflow and agent instruction drafting to use the same metadata extraction endpoint resolver as document metadata extraction.
  • Prevents draft workflow instructions from sending a selected multi-endpoint / Foundry-hosted metadata extraction deployment to the legacy global GPT/APIM endpoint.
  • Ref: resolve_metadata_extraction_client(), route_backend_agents.py, route_backend_workflows.py, WORKFLOW_INSTRUCTION_MODEL_ENDPOINT_FIX.md

Version and documentation

• Bumped application version to 0.261.061 in application/single_app/config.py.
• Updated release notes in docs/explanation/release_notes.md.
• Added/updated feature and fix documentation:
  • docs/explanation/features/CONTENT_SAFETY_BLOCK_THRESHOLD.md
  • docs/explanation/fixes/SAFETY_VIOLATION_METADATA_DISPLAY_FIX.md
  • docs/explanation/fixes/WORKFLOW_INSTRUCTION_MODEL_ENDPOINT_FIX.md
• Updated logging tag reference documentation.
• Updated admin documentation and settings field baseline coverage.

Validation

Passed:

git fetch upstream Development
git -c core.whitespace=blank-at-eol,blank-at-eof,space-before-tab,cr-at-eol --no-pager diff --check upstream/Development...HEAD
python -m py_compile <changed Python files>
python .\scripts\check_swagger_routes.py <changed route files>
python .\functional_tests\route_tests\test_route_blueprint_policy_inventory.py
python .\functional_tests\route_tests\test_route_unauthenticated_policy_contract.py
python .\functional_tests\route_tests\test_route_policy_test_coverage.py
python .\scripts\check_xss_sinks.py --base-sha upstream/Development --head-sha HEAD <changed app files>
python .\scripts\check_broken_access_control.py --base-sha upstream/Development --head-sha HEAD <changed Python app files>
python .\functional_tests\test_admin_review_record_lifecycle.py
python .\functional_tests\test_content_safety_violation_message.py
python .\functional_tests\test_tabular_model_http_diagnostics.py
python .\functional_tests\test_tabular_tool_error_retry_and_thoughts.py
python .\functional_tests\test_workflow_instruction_drafting.py
python .\functional_tests\test_docs_app_surface_coverage.py
python .\functional_tests\test_docs_site_quality.py
python .\ui_tests\test_admin_content_safety_threshold_ui.py
python .\ui_tests\test_admin_review_list_card_views.py
node --check .\application\single_app\static\js\admin\admin-safety-violations.js
node --check .\application\single_app\static\js\admin\admin_settings.js

[Chad Palmer (chadpalmer)]

Paul Lizer (@paullizer) Bionic711 This PR fixes a huge processing time (> 1-2 hours) for tabular files (excel for example) issue. Content safety slider for each content safety category have been added to the admin settings. The default is 4 to preserve previous behavior by default. Lastly, the draft workflow instructions button has been fixed to use the metadata extraction model setting rather than using the legacy global azure_openai_gpt_endpoint that can throw errors if it's not enabled.