Domain 3 – Integration (19%) – Highest weight

8 lessons, each ending with a practice scenario (Pro)

Log in to save your progress.

1. ✓3.1 — Capability Bloat and Least-Privilege ConfigurationLeast privilege = agent gets only the tools its role needs. Capability bloat = every tool beyond that. Removal beats mitigation every time.Scenario: Pro
2. ✓3.2 — Authentication, Authorisation, and the Gaps BetweenAuthentication = who is calling. Authorisation = what may they do. A security gap exists wherever effective privilege exceeds the requesting user’s legitimate need.Scenario: Pro
3. ✓3.3 — Accuracy–Latency Trade-offsEvery integration has a latency requirement (e.g. p95) and a quality requirement. The skill is holding both at once.Scenario: Pro
4. ✓3.4 — Observability and Monitoring at ScaleLLM systems break anecdote debugging: non-deterministic, so a single re-run proves nothing. A green uptime dashboard says nothing about answer quality.Scenario: Pro
5. ✓3.5 — RAG Pipeline Design: Chunking and IndexingRAG stages: ingest → chunk → embed → index → retrieve → (rerank) → generate. Generation quality is capped by retrieval quality.Scenario: Pro
6. ✓3.6 — Matching Retrieval to Data Shape and Query PatternRetrieval strategy rests on two axes: data shape and query pattern. Then name the failure that matters most and pick the cheapest strategy whose failure mode is acceptable.Scenario: Pro
7. ✓3.7 — Choosing the Integration MechanismFour mechanisms connect Claude to surrounding systems: MCP server, Direct API/SDK, Headless (SDK/CLI), Agent-to-agent.Scenario: Pro
8. ✓3.8 — Progressive Discovery vs Monolithic ContextTwo ways to load tools, procedures and knowledge: Monolithic — load everything upfront (simplest; best for small stable surfaces; cacheable); Progressive discovery — load each piece on demand (Skills, deferred tool loading, agentic retrieval).Scenario: Pro