Question 150
A project skill grows too large because it embeds all examples, policy manuals and implementation notes in one prompt. What structure is better?
- Convert all instructions into a hidden tool result
- Use a focused SKILL.md with referenced support files
- Remove the workflow steps and keep only a title
- Clone the whole document into every skill in the project
Explanation
Correct Answer
B:Separating core procedure from detailed resources reduces irrelevant context and eases updates. The skill should expose a clear entry point.
Incorrect Answers
A:Tool results are observations, not a durable skill definition. The procedure belongs in a maintained instruction file.
C:Without concrete instructions, the skill cannot consistently guide execution. Shorter should not mean empty.
D:Duplication increases maintenance and conflicting versions. Shared supporting resources are preferable.
Question 151
A repository uses both legacy custom slash commands and newer skills. Which migration practice is sound?
- Store migrated instructions only in a developer's private chat
- Preserve behavior when moving reusable commands to skills
- Delete all commands before examining their consumers
- Assume renaming a file creates identical behavior without checking syntax
Explanation
Correct Answer
B:Different mechanisms can coexist in a transition. Testing the resulting prompts and parameters prevents broken team automation.
Incorrect Answers
A:Personal conversation state is not a shared versioned replacement. The workflow should remain accessible to the team.
C:Other scripts and teammates may still invoke them. Abrupt removal can break repeatable workflows.
D:Instruction formats and argument handling can differ. Migration requires validating how the workflow is discovered and invoked.
Question 152
A code review process must be run on demand by a maintainer rather than on every conversation turn. What design is most suitable?
- Ask Claude to infer from every greeting that review is required
- Provide a named, explicitly invocable project skill or command
- Modify a pre-run hook to execute a release each time Claude starts
- Embed the full review in an unconditional global instruction
Explanation
Correct Answer
B:On-demand invocation separates a specialized workflow from general instructions. It avoids applying an expensive review procedure to unrelated tasks.
Incorrect Answers
A:Unrelated prompts do not authorize full repository checks. Explicit requests are clearer and safer.
C:A side-effectful unconditional hook is unrelated to on-demand review. It could cause serious workflow mistakes.
D:A global rule could trigger unnecessary work on trivial questions. The review belongs to an appropriate invocation boundary.
Question 153
A skill must produce a consistent handoff note to a second engineer. What should the author specify?
- Ask for an unbounded stream of internal model reasoning
- Required fields, evidence, status and output destination
- Only instruct the model to write something helpful
- Use the length of the note as the sole success metric
Explanation
Correct Answer
B:Clear output contracts improve repeatability and downstream consumption. The skill should distinguish complete from blocked work.
Incorrect Answers
A:Internal deliberation is neither a stable deliverable nor a suitable handoff contract. The result should be a concise artifact.
C:Vague goals do not establish the fields a colleague needs. Important evidence may be omitted.
D:Long notes can still omit the status or provenance. A content checklist is the relevant test.
Question 154
A developer makes a private experimental skill, but teammates should not receive it until approved. Where should it live initially?
- In the user-scoped skills directory, not the project
- Paste its complete content into an unrelated package's README
- Hide the skill in production build artifacts
- Commit the experiment immediately to the team's main branch
Explanation
Correct Answer
A:User-level skills support personal experiments. Moving it to shared project configuration can follow review and testing.
Incorrect Answers
B:An arbitrary README is not an explicit reusable skill definition. Discovery and invocation become unreliable.
C:Compiled outputs are not the normal authoring and discovery location for Claude Code skills. This complicates testing.
D:Shared configuration distributes unreviewed behavior. That conflicts with the desired trial isolation.
Question 155
An automated skill runs a checklist but sometimes reports pass without running the required tests. What should the team improve?
- Increase the number of examples while leaving the missing execution rule unresolved
- Ask Claude to use more enthusiastic final language
- Include explicit test commands, expected exit criteria and evidence requirements
- Remove failing tests from the checklist
Explanation
Correct Answer
C:A workflow should request observable results, not rely on a confident summary. Independent checks can reject a report missing test evidence.
Incorrect Answers
A:Examples may help phrasing but cannot substitute for the required test and acceptance gate.
B:Optimistic prose does not create actual test execution. The problem is missing verification steps.
D:Hiding failures makes release criteria weaker. The skill must enforce its documented quality conditions.
Question 156
A monorepo has Python services and TypeScript UI components. The lint guidance must apply only to files under ui. Which configuration best matches the requirement?
- Add UI lint commands to a global personal instruction for all repositories
- Put lint settings in an unrelated build log
- Ask the model to guess file ownership from extension alone each time
- Use a path-scoped Claude Code rule that matches ui files
Explanation
Correct Answer
D:Scoped rules place specialized instructions where relevant without burdening unrelated backend work. The match pattern should cover the intended paths.
Incorrect Answers
A:That applies beyond the UI subtree and beyond this project. The stated need is package-specific guidance.
B:A log is not a policy file that Claude Code reliably loads for the target path. The team needs a path rule.
C:Extensions can appear in many packages. An explicit path pattern is a more stable boundary.
Question 157
A repository's .claude/rules folder contains a rule targeting docs/**/*.md. When should the rule be relevant?
- Only for files named literally with asterisks
- Only after the documentation has been deployed
- For every binary and source file regardless of location
- When matching files trigger the configured path rule
Explanation
Correct Answer
D:Path patterns allow instructions to follow a file family. Correctly scoped guidance reduces accidental interference with other work.
Incorrect Answers
A:Glob syntax denotes pattern matching, not a filename containing stars. The rule covers matching descendants.
B:Rule loading relates to the working context, not runtime deployment. It guides local authoring operations.
C:A path-specific rule is designed to be selective. Treating it as global defeats the configuration intent.
Question 158
A rule for database migrations is triggered when editing unrelated frontend code because its path pattern is too broad. What should be fixed?
- Delete every project rule to avoid scope errors
- Restrict the path selector to the migration files
- Rename the frontend files to evade the pattern
- Add more migration text to reinforce its importance
Explanation
Correct Answer
B:The defect is selection scope, not content quality. A more precise pattern prevents irrelevant database instructions in UI work.
Incorrect Answers
A:Other targeted rules may be valuable. Fixing one broad selector is the smaller corrective action.
C:Changing application filenames to work around configuration errors is fragile and disruptive. The rule should be repaired.
D:Extra prose does not correct a rule that loads in the wrong context. It can increase irrelevant prompt overhead.
Question 159
Two path rules apply to the same file and require conflicting formatters. What should the maintainer do?
- Run both conflicting formatters alternately until the output stops changing
- Ignore the issue because path rules cannot overlap
- Assume Claude always knows which formatter the team prefers
- Resolve the overlap explicitly and define one authoritative formatter for that path
- Require each developer to remember which formatter takes precedence
Explanation
Correct Answer
D:Contradictory instructions create unpredictable assistant behavior. Rule ownership and pattern precedence should be reviewed for consistency.
Incorrect Answers
A:Competing tools may churn the file or produce repeated diffs. A single accepted standard is needed.
B:Globs and nested paths can match the same file. Overlaps require deliberate rule design.
C:The conflict exists in project guidance. Implicit preference cannot provide a reliable rule.
E:Manual recollection does not resolve contradictory project configuration. The rules need a single maintained source of truth.
Question 160
A team needs test generation conventions for files ending in .spec.ts inside packages/web. Which match strategy is best?
- Apply the rule to all files in every package
- Rely on the developer to restate the pattern with each prompt
- Use a pattern constrained to packages/web and the spec file suffix
- Apply the rule to every markdown file in the repository
Explanation
Correct Answer
C:Both directory and file type matter to the requirement. Scoped matching avoids applying the test rule to unrelated TypeScript files.
Incorrect Answers
A:An overly broad policy introduces noise and may conflict with non-web standards.
B:Manual reminders are less reliable than versioned path rules. The team wants reusable configuration.
D:Markdown documents are not the requested test files. The selector would miss its target.