3.1Beginner30 minutes
3.1 — Shrink a bloated support-agent toolkit
What you will learn
- Inventory tools by usage and blast radius
- Merge look-alike search tools
- Remove unused write APIs from a pilot
← Build Exercises
Domain 3
Integration
8 build exercises to practice the concepts in this domain.
3.2Intermediate40 minutes
3.2 — Close AuthZ on a refund tool path
What you will learn
- Separate AuthN from AuthZ on a sequence diagram
- Bind refunds to verified case context
- Prove denies with automated tests
3.3Intermediate35 minutes
3.3 — Decide a hot-path config under a p95 SLA
What you will learn
- Write dual accuracy and latency bars
- Compare two configs with quantified deltas
- Recommend ship, redesign, or waiver
3.4Advanced45 minutes
3.4 — Design LLM sampling and SLIs for high volume
What you will learn
- Define quality/cost/latency/tool SLIs
- Write a sampling and redaction policy
- Correlate traces to outcomes
3.5Intermediate45 minutes
3.5 — Structure-aware chunk and fresh index for policies
What you will learn
- Chunk policies on clause boundaries
- Attach provenance metadata
- Promote an index only after smoke tests
3.6Beginner30 minutes
3.6 — Route dense vs hybrid by query pattern
What you will learn
- Bucket queries into exact, semantic, and filtered
- Match sparse/hybrid vs dense
- Score recall per bucket
3.7Beginner25 minutes
3.7 — Choose MCP vs API/CLI vs agent-to-agent
What you will learn
- Score reuse, ownership, and isolation
- Pick a mechanism per integration
- Record rejected alternatives in an ADR
3.8Advanced40 minutes
3.8 — Progressive discovery plan for a monorepo assistant
What you will learn
- Reject dishonest monolithic dumps for multi-GB repos
- Design search → open → summarize with a facts pin
- Compare token use against a size-capped baseline