# Case Study: Adding a Feature — AI Coding Assistants: Use Them Well and Review Them Hard

Source: https://www.geekswithgeeks.com/en/ai-coding-assistants/wrap-feature-case

> Walk through adding a "reminders" feature with an assistant from request to merge.

## The workflow

You create a branch, share the relevant files and the project guide, and write the acceptance tests (or review ones the assistant drafts). You then ask for the implementation **in small steps**, running the tests after each. Before merging you read the whole diff using the checklist: do all new functions exist, are edge cases covered, are there string-built queries or leaked secrets, and did anything unrelated change? Finally you write the commit message yourself or edit the draft, and the pull request gets a normal human review.

## A full assisted feature

A disciplined workflow combines context, tests, small steps and careful review.

![Four stages: plan, test, build, review.](assets/figures/ai-coding-assistants/section-8-map.svg) — Figure 8.1 — Plan, test, build and review.

## The workflow on one page

Each line maps to a section of this course.

```text
1 git switch -c ai/add-reminders                      (Sec 6: isolate)
2 share files + project guide; state constraints         (Sec 2: context)
3 write / review tests first                             (Sec 2, 4: tests)
4 implement in small steps, run tests each time          (Sec 2, 3: steps)
5 review full diff with the checklist                    (Sec 5: review)
6 secrets / licence / policy check                       (Sec 6)
7 human PR review, then merge                            (Sec 7: teams)
```

**Quiz:** In the case study, when are the tests written?

- [ ] After merging
- [ ] Never
- [x] Before the implementation, so success is checkable
- [ ] Only if the build fails

*Answer:* Before the implementation, so success is checkable. Defining success first gives both you and an agent a clear target to verify against.