[Paul Lizer (paullizer)]

Summary

• Older public documents could disappear from chat source selection because the API ignored search and paging before truncating recent results. Apply filename/title filtering, allowlisted sorting, filtered totals, and pagination after canonical visibility and current-revision selection on both storage read paths.
• Report partial document/tag lookup failures with an explicit retry in shared Documents and # pickers, while retaining successful results and never widening the selected scope.
• Fix the public-only metadata extraction feature gate and persist independent metadata outcomes across personal, group, and public workspaces. Failed or empty analysis preserves indexed-content readiness and existing metadata; authorized users can retry without fabricating missing authors or bypassing execution-time guards.
• Align application-owned fixtures with the real API, add executable cross-scope and installed-SDK regressions, and document the behavior in version 0.261.347.

Existing populated metadata is preserved rather than regenerated wholesale. Existing records are not migrated or automatically retried. Configured providers, screening/publication restrictions, and access rules are unchanged.

Linked issue

Fixes #1778

Release Notes & Latest Features

• New Feature
• Bug Fix
• UI Enhancement
• Breaking Change
• Internal only

Is this visible to end users?

• Yes
• No

Is this admin-facing (Admin Settings, governance, deployment, config)?

• Yes
• No

Should this become a Latest Feature card?

• Yes
• No
• Already added

Screenshot needed for the card?

• Yes
• No
• Attached

Version bump

• application/single_app/config.py VERSION third segment bumped, or not needed because this is docs-only
• deployers/version.txt bumped, or not needed because deployers/ was not changed

Application: 0.261.346 -> 0.261.347. No deployer changes.

Testing / validation

Python commands used the isolated validation environment with repository-pinned dependencies. The complete Cosmos gate blocks network access.

• python -B scripts\check_cosmos_response_boundaries.py --run-contracts - passed: 1,088 tests and 244 subtests; 1,066 owners, 410 doubles, zero review gaps.
• python -B -m pytest -p no:cacheprovider -q --disable-warnings --tb=short --maxfail=1 functional_tests\test_public_chat_document_list_pending_artifacts.py functional_tests\test_workspace_metadata_outcomes.py functional_tests\test_document_auto_metadata_extraction_consistency.py functional_tests\test_multiselect_metadata_extraction.py functional_tests\test_content_screening_upload_only_intake.py - 227 passed.
• python -B -m pytest -p no:cacheprovider -q --disable-warnings --tb=short functional_tests\test_v2_public_chat_scope.py functional_tests\test_public_document_fixture_parity.py functional_tests\test_group_document_fixture_parity.py - 193 passed.
• python -B -m pytest -p no:cacheprovider -q --disable-warnings --tb=short --maxfail=2 functional_tests\test_public_document_management.py functional_tests\test_group_document_management.py -k 'metadata or worker or queue' - 88 passed.
• python -B -m pytest -p no:cacheprovider -q --disable-warnings --tb=short ui_tests\test_v2_chat_context_selection.py ui_tests\test_v2_public_chat_scope.py ui_tests\test_v2_public_documents.py ui_tests\test_v2_group_document_management.py - 202 passed, four pre-existing mobile-navigation failures. All four reproduce against unchanged HEAD frontend sources.
• python -B -m pytest -p no:cacheprovider -q --disable-warnings --tb=short ui_tests\test_v2_workspace_authoring.py -k knowledge_sources - 1 passed. Broader agent-authoring failures were also reproduced against the unchanged frontend and are not changed here.
• npm --prefix application\v2_ui run typecheck - passed.
• npm --prefix application\v2_ui run build - passed; existing bundle-size warning remains.
• python functional_tests\route_tests\test_route_blueprint_policy_inventory.py - 15/15 passed.
• python functional_tests\route_tests\test_route_unauthenticated_policy_contract.py - 10/10 passed.
• python functional_tests\route_tests\test_route_policy_test_coverage.py - 3/3 passed.
• python -B functional_tests\test_docs_app_surface_coverage.py - 7/7 passed; generated inventory is current.
• python -B functional_tests\test_docs_site_quality.py - 6/6 passed.
• git diff --check - passed.

No deployment or state-changing live validation was performed. The historical metadata-provider failure and effective deployed revision remain unconfirmed because the available Azure identity could not read the required processing records. Live source selection, metadata retry, and grounded answer/citation verification require a separately authorized workflow.

Documentation

• Release notes updated, or not needed
• Feature documentation updated, or not needed
• Fix documentation updated, or not needed

Added docs/explanation/fixes/WORKSPACE_DOCUMENT_DISCOVERY_AND_METADATA_FIX.md; updated public-workspace and document-management guides, chat controls, and existing metadata fix documentation. Release-note update was offered but not approved, so docs/explanation/release_notes.md is unchanged. No new feature page is required for this fix.

Security checklist

• New Flask routes include @swagger_route(security=get_auth_security())
• Settings sent to non-admin frontends use sanitize_settings_for_user()
• Browser JavaScript is served from local SimpleChat static assets only; no CDN-hosted JS
• No secrets, keys, connection strings, or local-only artifacts are included

No new Flask routes or browser-facing settings payloads were introduced. Scope, screening, feature gates, and queued-job authorization remain enforced.