teaching-knowledge-base-plan

name	teaching-knowledge-base-plan
description	Plan a teaching knowledge base for AI-agent support, including source types, metadata, RAG architecture, privacy boundaries, content governance, evaluation, and teacher-facing use cases.

Teaching Knowledge Base Plan

Use when the user asks to build a teaching knowledge base, RAG system, course knowledge base, AI tutor memory, or repository of active learning resources.

Scope

This skill plans the knowledge base. It does not install dependencies or implement a full system unless the user explicitly asks.

Use dissertation-knowledge-ops instead when the task is syncing, deduplicating, indexing, or deciding the source of record across research-wiki, knowledge-base, and Obsidian.

Source Types

Consider:

• active learning literature notes
• teaching strategy guides
• course outlines
• lesson plans
• assessment rubrics
• workshop outputs
• anonymized teacher concerns and needs
• institutional guidance
• AI-agent design requirements
• prototype evaluation notes

Workflow

1. Define users:
   • researcher
   • university teacher
   • AI agent
   • supervisor/examiner
2. Define use cases:
   • answer questions about active learning
   • suggest teaching activities
   • retrieve evidence for dissertation writing
   • support AI agent design decisions
   • compare adoption concerns across teachers
3. Specify information architecture:
   • documents
   • metadata
   • tags
   • evidence tables
   • concept pages
   • decision logs
4. Decide retrieval strategy:
   • simple Markdown wiki first
   • local search
   • vector search/RAG later
   • citations/source links required
5. Define privacy boundaries:
   • public teaching resources
   • restricted research notes
   • sensitive participant data
6. Define evaluation:
   • source citation accuracy
   • retrieval relevance
   • usefulness for teachers
   • hallucination rate
   • privacy compliance

Output

Create or update:

• knowledge-base/KB_PLAN.md
• knowledge-base/SOURCE_REGISTER.md
• knowledge-base/METADATA_SCHEMA.md
• knowledge-base/RAG_EVALUATION_PLAN.md

Starter Architecture

For a research-agent starter project, prefer:

1. Markdown knowledge base first.
2. Clear metadata and source registers.
3. Manual citation audit.
4. Local/private RAG only after the structure is stable.

Guardrails

Do not mix raw participant data with general teaching resources. Do not recommend uploading sensitive data to external services without explicit confirmation.