# TestBucks - the book's fictional coffee shop

TestBucks is a fictional coffee-shop training website inspired by familiar coffee-ordering experiences. It is not Starbucks and is not affiliated with Starbucks. The local app uses synthetic data and simulated payments only.

Published deployment URLs:

- Companion materials: https://howaitestssoftware.com/companion.html
- TestBucks: https://howaitestssoftware.com/testbucks.html
- Source ZIP: https://howaitestssoftware.com/testbucks-source.zip

For offline practice, extract the ZIP, open a terminal in the extracted `testbucks-lab` folder, and run `python3 -m http.server 3062`. Open `http://localhost:3062/` in your browser. This local address is only for your own copy.

The shared journey is: open the menu, find a vanilla latte, choose medium size and oat milk with an extra shot, add one to the cart, verify $6.60 including training tax, and complete a simulated pickup order. No real payment provider is connected.

The baseline app supports menu search, drink customization, cart changes, BREW10, simulated success/decline/timeout, order history, and reset. Data stays in this browser. Advanced book examples involving accounts, server ledgers, native apps, distributed services, and malicious note rendering are illustrative extensions, not features already implemented by this single-file lab.

## Reproducible first defect

Open https://howaitestssoftware.com/testbucks.html?fault=search#menu and search for `vanilla`. The seeded search fault hides the vanilla latte even though it is in the catalog. Remove `?fault=search` and repeat as the negative control. For a local copy, add `?fault=search#menu` to its address. This deliberately enabled failure is a learning fixture, not a finding against a real coffee company.

## Evidence and interpretation

The book's TestBucks screenshots are captured from this local app. Agent transcripts, scores, network excerpts, and advanced scenarios are illustrative unless an accompanying run record explicitly says otherwise. Original Bing and TaskBoard historical evidence remains archived; it is not relabeled as TestBucks evidence.

The circular TestBucks bug badge is `testbucks-logo.png`, recreated from the
existing badge on `testbucks-cup.png`. Keep both images beside `index.html`.
Do not substitute a standalone checkmark for the brand logo. Refresh the full-page
book captures with `node scripts/capture_testbucks_branding.cjs` from the project
root, with Playwright available to Node.
# Book screenshot scenarios

The book's compact screenshot flows are reproducible teaching examples. The normal
shop is still `index.html`. Use `book-scenarios.html?case=adaptive` for the separate
variants; run `python3 -m http.server 8080` in this folder and open the local page.

Available `case` values:

- `adaptive`: promotion, changed customization flow, and the same latte intent.
- `silent`: a false confirmation with no saved order.
- `recovery`: timeout after mock acceptance, then an idempotent retry.
- `diagnostics`: vague error versus controlled diagnostic evidence.
- `signin`: simulated identity change, layout overflow, and a responsive fix.
- `personalized`: a synthetic profile and a simulated voice transcript.

All payments and identities are simulated. The recovery ledger is in memory and
resets on reload; it is not a database or payment gateway. The sign-in exercise
does not implement authentication. The voice exercise uses a fixed interpretation
of the example transcript, not speech recognition or an LLM. Navigation buttons
outside each scenario return to the starter app; reload the scenario URL to reset.