Skip to content

FAQ Coverage Gaps

Concepts from the learning graph not yet covered in docs/faq.md. Organized by priority based on centrality in the learning graph.

Generated: 2026-05-19 | Total concepts: 496 | Covered in FAQ: ~108 | Gaps: ~388

Critical Gaps (High Priority)

High-centrality concepts with many dependencies that should be addressed in FAQ:

Topic 9 — Decision Traces (gaps)

  1. Trace Completeness — high centrality; many chapters reference it
  2. Suggested: "How do you ensure a decision trace is complete?"
  3. Policy Version Reference — foundational to compliance chapter
  4. Suggested: "How are policy versions tracked in decision traces?"
  5. Trace Replay — key for audit and debugging
  6. Suggested: "What is trace replay and when do you need it?"
  7. Counterfactual Trace — advanced but high-value audit concept (partially covered)
  8. Real-Time Trace Recording — architecture decision with high impact
  9. Suggested: "How do you capture decision traces in real time without adding latency?"

Topic 12 — Integrating LLMs (gaps)

  1. Function Calling Pattern — widely used in chapters 14 and 16
  2. Suggested: "What is the function calling pattern for context graph retrieval?"
  3. Structured Output with Context — important for reliable agent outputs
  4. Suggested: "How do you get structured JSON outputs from an LLM using context graph data?"
  5. Prompt Engineering with Context — high practical relevance
  6. Suggested: "What prompt engineering patterns work best with context graph retrieval?"
  7. Few-Shot Context Injection — Bloom's Apply level, practical
  8. Suggested: "What is few-shot context injection and how does it use decision traces?"

Topic 14 — AI Agent Architecture (gaps)

  1. ReAct Pattern — fundamental agent loop design
    • Suggested: "What is the ReAct pattern and how does it apply to context graph agents?"
  2. Human-in-the-Loop — critical for graduated autonomy
    • Suggested: "When and how should humans remain in the loop in context graph workflows?"
  3. Plan-and-Execute Pattern — widely referenced agent design
    • Suggested: "What is the plan-and-execute pattern for context graph agents?"
  4. Agent Trace — distinct from decision trace, important for debugging
    • Suggested: "What is an agent trace and how does it differ from a decision trace?"

Cross-Cutting — LLM Foundations (gaps)

  1. Prompt Injection Risk — security concern directly relevant to context graph inputs
    • Suggested: "What is prompt injection and how does it threaten context graph systems?"
  2. Context Window Limit — foundational LLM constraint
    • Suggested: "What is the context window limit and how does it constrain context graph design?"
  3. In-Context Learning — key mechanism underlying context graph effectiveness
    • Suggested: "What is in-context learning and why does it make context graphs effective?"

Medium Priority Gaps

Topic 15 — Enterprise Use Cases

  1. Finance Automation Use Case — concrete example for business readers
  2. Sales Engagement Use Case — high-value for sales-tool builders
  3. Engineering Incident Use Case — popular in developer audiences
  4. Legal Compliance Use Case — important for regulated industries
  5. ARR Definition Conflict — specific, memorable example of the context problem

Topic 17 — Market Strategy

  1. Competitive Moat — strategic concept important for founders
  2. Glue Function — the "beachhead signal" concept is mentioned but not named precisely
  3. Enterprise AI Market — macroeconomic context for the book's thesis

Topic 13 — Graph Data Modeling

  1. Subgraph Extraction — key retrieval primitive
  2. Context Graph Migration — practical concern for evolving deployments
  3. Graph Constraint Enforcement — schema integrity in production
  4. Event-Driven Graph Update — real-time ingestion pattern

Topic 6 — Process Mining

  1. Process Discovery — foundational process mining step
  2. Conformance Checking — important for detecting workflow deviations
  3. Append-Only Log — architectural choice with implications for audit
  4. Event Sourcing — related pattern, important for context-graph ingestion design

Cross-Cutting — Graph Theory

  1. Centrality Measure — used for identifying high-priority concepts and entities
  2. Community Detection — useful for surfacing related decision clusters
  3. Subgraph Matching — important for pattern-based precedent retrieval
  4. Knowledge Graph Embedding — connects graph and vector search chapters

Low Priority Gaps

Leaf nodes and highly specialized concepts appropriate for future FAQ updates:

Topic 1 — Knowledge Graphs and LPGs

  1. GraphML, GraphSON serialization formats
  2. Open World vs Closed World Assumption
  3. Graph Index (implementation detail)

Topic 5 — Metadata Registries

  1. UMLS (domain-specific medical standard)
  2. NIEM (government data standard)
  3. Dublin Core Metadata
  4. Administered Item (ISO 11179 detail)

Topic 6 — Process Mining

  1. CQRS Pattern
  2. IEEE XES attribute structure (partially covered)
  1. HNSW Index (implementation detail)
  2. Product Quantization
  3. BM25 (sparse retrieval)
  4. Approximate Nearest Neighbor

Cross-Cutting — Security

  1. Differential Privacy
  2. Federated Learning
  3. AI Red Teaming

Suggested Additional Questions (Top 10)

Based on the critical gaps above, the highest-value additions to docs/faq.md are:

  1. "What is the ReAct pattern and how do context graph agents use it?" (Core Concepts)
  2. "How do you ensure a decision trace is complete?" (Technical Details)
  3. "What is prompt injection and how does it threaten context graph systems?" (Common Challenges)
  4. "When and how should humans stay in the loop in context graph workflows?" (Best Practices)
  5. "How do you capture decision traces in real time without adding latency?" (Technical Details)
  6. "What is the function calling pattern for context graph tool use?" (Technical Details)
  7. "What is in-context learning and why does it make context graphs effective?" (Core Concepts)
  8. "How do you handle a finance exception workflow with context graphs?" (Best Practices)
  9. "What is a competitive moat in the context graph market?" (Advanced Topics)
  10. "What prompt patterns work best when injecting context graph data?" (Best Practices)