Study the context, then drill it
Scenarios
Claude Certified Architect – Foundations publishes 6 scenarios in advance and anchors questions to them. Read each one properly before you answer anything about it — the context is the part you can prepare with certainty.
6 scenarios are published; an assembled set draws from 4 of them. Which 4 is not knowable in advance, so the only safe preparation is to know all 6.
Published scenarios
6 of 6 readable from today's verified catalog
-
Scenario
13 practice items
Claude Code for Continuous Integration
Claude Code runs inside a CI/CD pipeline, reviewing pull requests, generating tests and posting feedback. The design problem is producing actionable findings without a false-positive rate that costs developer trust.
Insufficient evidence 0 graded answers · 0 of 13 items met
-
Scenario
21 practice items
Code Generation with Claude Code
A team adopts Claude Code for generation, refactoring, debugging and documentation, and has to fit it to an existing workflow: shared configuration, custom commands, and knowing when planning beats going straight to the edit.
Insufficient evidence 0 graded answers · 0 of 21 items met
-
Scenario
22 practice items
Customer Support Resolution Agent
An Agent SDK support agent handles ambiguous requests - returns, billing disputes, account problems - through custom MCP tools that reach backend systems. It must resolve most cases on first contact while recognising the ones it should hand to a human.
Insufficient evidence 0 graded answers · 0 of 22 items met
-
Scenario
20 practice items
Developer Productivity with Claude
Agent SDK tooling helps engineers navigate unfamiliar and legacy codebases, generate boilerplate and automate repetitive work, combining the built-in file and search tools with MCP servers.
Insufficient evidence 0 graded answers · 0 of 20 items met
-
Scenario
20 practice items
Multi-Agent Research System
A coordinator agent delegates to specialised subagents that search, analyse documents, synthesise findings and write reports. The system has to produce comprehensive, cited output, which makes decomposition, context passing and provenance the failure points.
Insufficient evidence 0 graded answers · 0 of 20 items met
-
Scenario
16 practice items
Structured Data Extraction
A pipeline extracts fields from unstructured documents, validates them against JSON schemas and feeds downstream systems. Accuracy has to hold across varied document formats and degrade gracefully on edge cases.
Insufficient evidence 0 graded answers · 0 of 16 items met
Coverage counts your own already-graded answers against the items today's catalog anchors to each scenario. It says how much of this material you have met, and nothing about how an official exam attempt would go — Google publishes no score mapping and this platform cannot guarantee success on the official exam.