CCAR-F - Claude Code Configuration & Workflows (20% of the exam) - Section 3.1
Configure CLAUDE.md files with appropriate hierarchy, scoping, and modular organisation.
The configuration hierarchy runs user-level, project-level and directory-level, and user-level instructions are never shared with teammates through version control. Candidates should diagnose a hierarchy fault such as a new team member missing instructions that live in someone's personal configuration, use import syntax to pull package-relevant standards files in, split a large file into topic-specific rules, and verify which memory files actually loaded.
CLAUDE.md hierarchyuser versus project scopeimport syntax for modularitytopic-specific rule filesmemory command for loaded files
A contractor joining the Meridian Platform team notices that Claude Code consistently renames generated helpers with a prefix that appears in none of the files he can find in the checkout, and neither the root CLAUDE.md nor the package level memory files mention it. Before raising a defect he runs the memory command in his session. What does running it establish?
AIt lists the memory files in scope and reconciles them, rewriting any rule that conflicts across two scopes so that the project file and the user file agree from the next turn onwards.
BIt prints the instructions the model actually followed on recent turns, so the prefix can be attributed to the turn that applied it rather than to any particular memory file held on disk.
CIt opens the memory files loaded for the session at each scope, so he can see whether the prefix rule arrived from a user level file on this machine rather than from the repository.check_circle Correct
DIt reloads every CLAUDE.md found beneath the working directory, so any subtree file that had not yet been triggered is added to the session and becomes available for inspection.
The memory command reveals which memory files are loaded at which scope, separating personal machine local rules from committed project rules. Unexplained behaviour usually means a rule is in play from a scope the person searching has not looked at, and a repository search covers only one of those scopes. Listing the loaded memory files by scope answers the question directly, because a rule that shows up under user scope is local to that machine and a rule under project scope is shared. That distinction is what decides whether the fix is a personal edit or a commit.
Why A is wrong: Automatic reconciliation would be a useful feature and the command does deal in scopes, which makes this attractive. It is wrong because the command surfaces the files for a human to edit and does not rewrite rules on its own.
Why B is wrong: Per turn attribution is exactly what he wants, so this is a tempting reading. It is wrong because the command reports loaded memory rather than a trace of which instruction drove which turn.
Why C is correct: Correct. The command makes the loaded set visible and editable per scope, which is the direct way to distinguish a personal machine local rule from a committed project rule.
Why D is wrong: Forcing subtree memory into context sounds like a plausible companion feature. It is wrong because the command is about viewing and editing what is loaded, not about pre-loading files that on demand scoping has deliberately left out.
Exam traps in Claude Code Configuration & Workflows
Answers that look right on this material and are not. Each one is a distractor from a different question in the CCAR-F bank for this domain.
Copy the mocking rule back into the root CLAUDE.md as well, so it is loaded regardless of whether the imported topic file is reachable from the root file.
errorWhy it is wrong: Tempting because it would very likely restore the behaviour, but it leaves a broken import chain in place, duplicates a rule that will now drift, and treats a symptom without establishing the cause.
It applies to every project on her machine and is folded into the repository at her next commit, so colleagues pick the rule up as soon as they pull the branch she is working on.
errorWhy it is wrong: Sharing a convention by committing it is the right instinct, which makes this plausible. It is wrong because a user scope file lives outside the checkout and no commit sweeps it in; the rule has to be written into the project file deliberately.
Ask each of the other engineers to copy the rule into the CLAUDE.md in their own home directory, so the same wording is present on every machine the team works from.
errorWhy it is wrong: Tempting because it does put the rule in front of every session, but user memory is per-machine and uncommitted, so the copies drift, new joiners start without it, and nothing in review catches a machine that lacks it.