learning

---

name: learning description: Updates local AgentDB vector store with new case patterns while monitoring for bias introduction license: MIT compatibility: opencode metadata: audience: developers workflow: clinical-pipeline

What I do

I update the vector store with the new case pattern for future RAG retrieval. I monitor the update for potential bias introduction and ensure diverse demographic representation in stored patterns.

When to use me

Use this when:

• Recommendations are generated and you need to update the knowledge base
• You're storing this case for future similar case retrieval
• You need to ensure the vector store remains bias-free

Key Concepts

• Vector Store Update: Add new embedding to AgentDB
• Bias Monitoring: Check for demographic skew in stored patterns
• Case Pattern: Embedding + metadata for future retrieval
• learning_updated: State flag after update complete

Source Files

• services/agentDB.ts: Vector store operations
• types.ts: ReasoningPattern interface

Code Patterns

• Generate embedding for the current case
• Store with demographic metadata and outcomes
• Validate demographic diversity in storage

Operational Constraints

• Must maintain demographic diversity in vector store
• Monitor for bias drift over time
• Log all updates for audit trail