“Looks good” is easy to write under a short code change. A useful beginner review explains what was checked, connects feedback to the intended behavior, and gives the author a clear next action.
For a coding bootcamp or tutoring group, start with one small diff: the lines removed and added in a proposed change. Give learners a written requirement, a few examples to reason through, and time to respond to each other's feedback. Focus Room can provide the browser website rules around the session; your repository, editor, runtime, and calling tools handle the coding work.
Give the reviewer a requirement before the code
Use a fictional training exercise with an author and a reviewer. The author proposes a change to countCompleted(tasks). The reviewer receives this contract:
- The input is an array of ordinary task objects, each with a boolean
completedfield. - Count only tasks whose
completedvalue istrue. - Exclude tasks marked
false. - An empty array returns
0.
Keep the exercise within that input contract. Handling missing fields, strings, or invalid inputs would require another decision; learners do not need to invent those requirements during this review.
Share the requirement, the before-and-after change, and the author's existing test example. Confirm that both people can open the materials before starting. In a pair, have the reviewer read independently before the author explains the implementation. In one-to-one tutoring, the tutor can take either role.
GitHub's pull-request review quickstart describes reading the summary and context, commenting on changed lines, and submitting feedback. You can practice the same conversation in a course document before learners need a real repository workflow.
Read one small diff against the contract
This is the proposed change for the fictional task counter. A minus sign marks a removed line; plus signs mark added lines. They are diff markers, not part of the JavaScript to run.
function countCompleted(tasks) {
- return tasks.length;
+ return tasks.filter(task =>
+ task.completed === true
+ ).length;
}
The old function counts every task. The proposal filters for completed === true and counts the matching objects. Ask the reviewer to explain that behavioral change before discussing names or formatting.
Start with three predictions:
- Three tasks marked
true,false,trueshould produce 2. - An empty array should produce 0.
- Two tasks both marked
falseshould produce 0.
The proposed implementation meets these examples. There is no need to invent an error so the reviewer has something critical to say. Checking a correct change and explaining why it fits the requirement is useful practice too.

Find what the existing example does not show
Suppose the author has supplied only this test case: two tasks marked true return 2. Both the old function and the proposed function pass that example. It checks an all-completed input without checking whether incomplete tasks are excluded.
Ask learners to choose a case that distinguishes the behaviors. The mixed input does: the old function returns 3, while the proposal returns 2. The all-false input also distinguishes them: the old function returns 2, while the proposal returns 0. The empty case checks the stated boundary, though both versions return 0 there.
Run these checks in the prepared JavaScript environment if the lesson includes execution. Have the learner predict each result first and keep a note of the actual result. If learners are only tracing code, label the results as predictions rather than claiming the tests ran.
Keep the review focused on the patch and the contract. A discussion about a new framework, a different data model, or elaborate validation can wait for a separate task.
Write one comment the author can act on
A review comment should identify the relevant code or behavior, explain the concern, and ask for a specific follow-up. For this exercise, the concern is the limited test example, not a demonstrated implementation failure.
A model comment could read:
The filter matches the requirement to count only completed tasks. The current all-true example would also pass with the old implementation. Could you add a mixed true/false case and an all-false case to check exclusion, plus the required empty-array case?
This gives the author a reason for the request and concrete cases to add. “Needs better tests” leaves them to guess what is missing. Avoid turning a personal preference into a correctness claim; explain whether the comment concerns required behavior, evidence, or optional style.
If learners are using GitHub, remind them that pending review comments remain private until the review is submitted. Otherwise, agree where the feedback will be shared in the lesson tools. Focus Room does not submit the review for them.

Ask the author to respond, then review again
The author should address the request with evidence rather than a bare “done.” A fictional response after actually running the added checks might be:
Added mixed, all-false, and empty-array checks. The mixed input returns 2, and the other two return 0. The implementation is unchanged because it already filters for true values.
Have the reviewer inspect the added cases and compare their assertions with the contract. Do they contain the promised inputs? Do they expect the right counts? Distinguish seeing a passing test report from independently executing the tests.
Finish with a short summary of what was checked and any remaining question. For this exercise, the summary can say that the boolean-input contract and listed cases were reviewed; invalid-input handling was outside the task. A lesson review does not authorize merging into a real project. Follow that project's normal permissions and review process.
If the task instead begins with a reproducible failure, use the beginner debugging session format. That lesson has a different starting point from reviewing this proposal.
Set a clear browser boundary for the session
For Focus Room, allow the course, repository or review page, meeting, and approved reference hosts the task needs. Test the complete route, including sign-in and links that open on another hostname. Learners can use the student session guide for joining help.
Focus Room matches allowed hostnames and their subdomains in the participating browser while the extension is active. Allowing a repository host permits more than one repository page; it does not grant repository access or restrict a learner to a single pull request.
Focus Room does not read or run code, inspect the learner's screen, host pull requests, run tests, create automated reviews, or merge changes. It does not control other browsers, devices, or desktop editors. Review evidence comes from the code, the learner's explanation, and the tools used for the task.
To try the routine, create a Focus Room account, prepare one small change and its requirement, and test the resource route before inviting the group. End the session when the review work finishes.

