# XJobsFinder — Pilot Test Pack
*Standardized manual test scripts — one per flow. For the testers (Oli, Oscar) and the pilot group.*

Everyone runs the **same** scripts the same way, reports the **same** way, and each script has a **companion video** you can watch first, then follow along. Findings all land in one place (**SUPER QA**), deduped.

---

## ▶ Watch first — Platform Overview (~2 min)
Before any flow, watch the **Platform Overview** — a 2-minute narrated tour of the whole platform (résumé → jobs → exchange → interview), so the pieces make sense before you test them. Script: `00-Platform-Overview-video.md` · **▶ [Watch the Platform Overview](https://www.loom.com/share/b8fc206038dc4988b420c742b91db513)**

## The flows (one script each)
| # | Flow | Script | Video | ~Time |
|---|------|--------|-------|-------|
| 0 | **▶ Platform Overview** — the whole platform in 2 min (watch first) | `00-Platform-Overview-video.md` | [▶ Watch](https://www.loom.com/share/b8fc206038dc4988b420c742b91db513) | 2 min |
| 🏁 | **Full Journey** — the whole pipeline in ONE session: ATS → Match → Customize → Interview | `00-Full-Journey.md` | _link_ | 20 min |
| 1 | **Auth & Recovery** — sign up, log in, forgot password | `01-Auth-and-Recovery.md` | _link_ | 8 min |
| 2 | **Resume · Fix it (ATS)** — optimize + the paywall | `02-Resume-Fix-ATS.md` | _link_ | 12 min |
| 3 | **Resume · Build it (AVA)** — build from scratch | `03-Resume-Build-AVA.md` | _link_ | 10 min |
| 4 | **Jobs · Find** — Personal emails (Gmail/Yahoo) + Job boards | `04-Jobs-Find.md` | _link_ | 8 min |
| 5 | **Match · Quick Match → Job Exchange** | `05-Match-and-Exchange.md` | _link_ | 12 min |
| 6 | **Land · Interview** — SME / Coach / Trainer | `06-Land-Interview.md` | _link_ | 12 min |
| 7 | **Payments · Paywall & Upgrade** | `07-Payments.md` | _link_ | 8 min |

Full pass ≈ 70 min · Essential-only (⭐ steps) ≈ 25 min.

---

## Before you start (every script)
1. **Fresh Incognito / Private window.** No prior XJobs tabs open. (Clears any saved session.)
2. **A brand-new account each run** — sign up with a **Gmail +alias**: `yourname+p1@gmail.com`, then `+p2`, `+p3`… (each is a clean new user; disposable domains like `example.com` are rejected).
3. **Open DevTools → Console** (F12) so you can see errors. Red errors = report them.
4. **URL:** `https://xjobs-final-production.up.railway.app` (or the link Jorge gives you).
5. **Have ready:** a résumé as **.docx**, a résumé as **PDF**, any **non-résumé** document (a report/letter), and a **job description** you can paste.

**Browser assignments** (catch platform-specific bugs):
- **Oli → Chrome** · **Oscar → Firefox** · **Pilot group → whatever you normally use** (tell us which).

---

## How to report (the only rule that matters)
On **every screen** there's a 🐞 **"Found a bug?"** button (bottom-right). The moment something looks wrong, confusing, or just off:
1. Click 🐞 → **talk (🎤) or type** what you saw / what you were trying to do.
2. It silently grabs a **screenshot** of exactly what's on your screen and files it.
3. Keep going — report as many as you like. Everything rolls up to **SUPER QA**, deduped.

For a structured run (testers), you can also use **`Submit a run`** on the dashboard with: **step # · browser · expected · actual · screenshot**.

⭐ = **P0 (blocking).** If a ⭐ step fails, report it **immediately**.

---

## The standard (so every script looks the same)
Each flow script follows this shape:

```
# Flow N — <Name>      ·  ~time  ·  browser
> One line: what this flow is and why it matters.

## Preconditions
- [ ] P1 … P2 …  (fresh state + what to have ready)

## Stage 1 — <name>
| # | Action | Expected result |
|---|--------|-----------------|
| 1 ⭐ | do this | see this |
| | **RED FLAG** | the failure mode to watch for |

## Report
🐞 Sidekick on any screen · or Submit a run: step # · browser · expected · actual · screenshot.
```

- **Stages** group the steps logically. **Numbered steps** are one action + one expected result.
- **⭐** marks P0 (must-pass) steps. **RED FLAG** rows name the common failure so testers know what "broken" looks like.
- Keep expectations **concrete** (exact text / behavior), so anyone can judge pass/fail without guessing.

---

## Videos
Each flow has a short companion video (≤3 min): watch it once to see the happy path, then run the script yourself and report any difference. Video **scripts** (what to record / say) live next to each test script (`*-video.md`) so they can be re-recorded as the product changes. Drop the recorded video link into the table above.

### 🎥 How to record one (Loom)
1. **Loom desktop app** → **"Screen Only"** (no camera needed). *(Or the Chrome extension → "Current Tab".)*
2. ⚠️ **Turn ON system/tab audio** — this is the #1 miss. The agent's voice (AVA) only records if sound capture is on (desktop: system audio; extension: ✅ "Share tab audio").
3. **5-second audio test first:** record 5 sec, trigger one spoken line, stop, **play it back.** Hear the voice → do the real take. Silent → fix the audio toggle and retry. (Saves you a perfect silent recording.)
4. Record → follow the flow's `*-video.md` → **Stop.** Keep each clip **under 5 min** (Loom's cap).
5. Loom **auto-uploads to your Loom cloud** and gives you a **share link** (`loom.com/share/…`) — the video lives there, not as a file on your Mac.
6. **Paste that link** into the matching row of the table above (and the `🎥 Video` line of the flow's script).

**Stop signals (so the clip doesn't run long):** the Platform Overview shows a green **"✓ Demo complete — you can stop the recording"** banner when AVA finishes. For flow videos, stop after the last narrated beat in the `*-video.md`.

---

## Where findings go
🐞 Sidekick + `Submit a run` → `qa_results` (Postgres) → **🦸 SUPER QA** tab on `/dashboard` (click **Consolidate** to build the deduped, prioritized list, each with its screenshot). Same bug from 3 testers → **1** item.