Claude Code Configuration & Workflows
Always-loaded, summoned on demand, or no-human-in-the-loop — sort every capability into one.
How you configure and steer an AI agent that works inside a code repository. The single most important instinct: know what is ALWAYS LOADED (the project memory file) versus SUMMONED ON DEMAND (skills, subagents, workflows) versus what runs with NO HUMAN IN THE LOOP (CI/CD). Almost every question here is really asking you to sort a capability into one of those three buckets, then apply least privilege to tools and read credentials from named secrets.
6 task statements on the blueprint
- 3.1Configure CLAUDE.md files with appropriate hierarchy, scoping, and modular organization4 objectives · 4 exam traps · 2 primary sources · 18 practice questions
- 3.2Create and configure custom slash commands and skills4 objectives · 5 exam traps · 3 primary sources · 13 practice questions
- 3.3Apply path-specific rules for conditional convention loading3 objectives · 4 exam traps · 2 primary sources · 6 practice questions
- 3.4Determine when to use plan mode vs direct execution4 objectives · 4 exam traps · 2 primary sources · 6 practice questions
- 3.5Apply iterative refinement techniques for progressive improvement4 objectives · 5 exam traps · 2 primary sources · 5 practice questions
- 3.6Integrate Claude Code into CI/CD pipelines5 objectives · 5 exam traps · 3 primary sources · 11 practice questions
The other domains
Working the whole blueprint? The interactive study guide carries the concept cards, the mock quiz, the audio lessons and the practice simulation.