Session Objective

Focused implementation-reviewer pass on the current Salesforce and Litify Agentforce work, with no file edits.

What Was Reviewed

• AttorneyHubInsuranceKnowledgeAction and test
• AttorneyHubNickTeamResourcesController and test
• attorneyHubNickTeamResources LWC
• Attorney_Hub_Insurance_Knowledge_Search GenAI function schema
• Attorney_Hubs_Access permission set addition
• Generated attorney aiAuthoringBundles

Live Checks

• Confirmed LITIFY_ORG connected.
• Confirmed a Litify matter query worked.
• Counted linked insurance rows: 58,714 total, 4,265 linked to open matters, 4,076 for Nick by attorney-name query, 541 open Nick rows, 2,972 UIM-type rows.
• Confirmed the carrier formula field can be used in a SOQL LIKE filter with a Progressive sample query.

Findings

1. High: The insurance action only queries the latest 700 candidate insurance rows before applying carrier, adjuster, status, coverage, missing-coverage, and matter-status filters in Apex. This can return false zero-result answers on real questions because the org has far more matching candidate rows than the pre-filter cap.
2. High: Org-wide insurance search is exposed through the LWC checkbox and generated Agentforce instructions without a separate leadership/custom-permission gate.
3. Medium: The Apex class uses with sharing, but it does not enforce object-level or field-level read security before returning insurance, claim, policy, limits, adjuster, and matter context.
4. Medium: The tests validate deployment and helpers, but they do not create representative insurance rows or assert filter correctness through the real query path.

Handoff