Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
64 changes: 52 additions & 12 deletions docs/user-guide.md
Original file line number Diff line number Diff line change
Expand Up @@ -95,6 +95,12 @@ A trainee's home shows what's available and their progress. The left nav groups

### The training menus

**My Assignments** — the exercises a trainer (or the system) has assigned to you,
with their due dates and your attempts. Two badges matter here: **Remediasi** marks
a targeted exercise auto-assigned after you failed something, and **Tes ulang** marks
a follow-up retest to confirm the lesson stuck (see *The remediation loop* below).
Start an assignment from this page.

**Scenarios** — branching, consequence-driven simulations. You read a message,
choose how to respond, and your decisions move a susceptibility score. There is
no "obviously correct" option; the server holds the score and the branches.
Expand All @@ -115,22 +121,33 @@ You get a per-criterion breakdown and coaching. (Needs an LLM configured.)

![Write Your Reply](images/04-write-reply.jpg)

**AI Challenge** — open-ended roleplay against an adversarial persona. Hold the
line in a live chat; leaking a secret or agreeing to the demand is a compromise,
resisting under escalating pressure is a win. The outcome and score are decided
server-side from what you type. (Needs an LLM configured.) This is no longer a
headline menu — it is reached when a trainer **assigns** it, so it works as a
targeted practice / remediation exercise rather than self-serve browsing.

![AI Challenge](images/05-ai-challenge.jpg)

**Quizzes** — multiple-choice knowledge checks with instant feedback. Graded on
the server; the answer key is never sent to the browser.

![Quizzes](images/07-quizzes.jpg)

**Training Campaigns** — multi-stage practice chains (email → call → social →
extraction) that unlock stage by stage.
> **Assigned-only exercises.** Two more exercise types exist but are *not* self-serve
> menus — they appear only when a trainer assigns them (and show up in **My
> Assignments**): **AI Challenge** (open-ended roleplay against an adversarial
> persona — leaking a secret or agreeing to the demand loses, resisting wins; scored
> server-side; needs an LLM) and **Training Campaigns** (multi-stage practice chains:
> email → call → social → extraction).

### The remediation loop

Pretexta is a closed loop — **measure → remediate → retest** — and you experience it
in *My Assignments*:

1. **You fail something.** A click in a live phishing simulation, or a low score on a
lab exercise.
2. **Targeted practice is assigned automatically.** The server looks at *which*
Cialdini principle you fell for (authority, scarcity, …) and assigns a focused
exercise on exactly that — tagged **Remediasi**.
3. **You pass it, then retest.** When you clear the remediation, a variant **Tes
ulang** is scheduled a little later to confirm the lesson held.

You do not request any of this; it is created for you. A trainer sees the same loop
across their cohort in Assessment Reports.

### Progress

Expand Down Expand Up @@ -190,6 +207,28 @@ stays manual by design (G8) — it is not the headline.
- **Luncurkan ke titik rawan** — from the Attack Surface Map, one click pre-fills a
campaign with the most-susceptible, emailable people.

### Laporan Phishing (Report-Phish inbox)

The defensive mirror of a campaign — **reporting a phish is the behaviour you want**,
so Pretexta measures it. When someone forwards a suspicious email to the abuse
mailbox, the server classifies it and the page shows each report with a verdict:

- **Simulasi tertangkap** — it was one of your own simulations; the reporter is
**credited automatically**, which flows straight into that campaign's report rate.
- **Perlu ditinjau** — not one of yours; click **Tandai ancaman** or **Tandai aman**
to triage it.

A **top-reporters** leaderboard surfaces who defends actively. In production a mail
rule or poller feeds the inbox; the **Catat laporan** form lets you record one by
hand for a demo. (Integration endpoint: `POST /report-inbox/ingest`.)

### Analitik Live

Report-rate **trend across campaigns** (oldest → newest), a per-campaign comparison,
and **repeat offenders** — the people who clicked or submitted in more than one
campaign. Those are exactly the people the remediation loop should be hardening, so
this is where you confirm training is actually moving behaviour over time.

---

## 5b. Human Attack Surface Map (admin / trainer)
Expand Down Expand Up @@ -240,12 +279,13 @@ only revealed on a node click — to the same staff who already see risk profile
| Assignments | Command | trainer/admin | Assign exercises |
| Assessment Reports | Command | trainer/admin | Trainee results & risk |
| Riwayat User | Command | trainer/admin | Per-user history |
| My Assignments | Training | all (trainee) | Assigned exercises; Remediasi / Tes ulang badges |
| Scenarios | Training | all | Branching simulations |
| Inbox Triage | Training | all | Signal-detection exercise |
| Write Your Reply | Training | all | LLM-graded free-text |
| AI Challenge | (assignment only) | all | Persona roleplay — launched from an assignment, not a menu |
| Quizzes | Training | all | Knowledge checks |
| Training Campaigns | Training | all | Multi-stage practice |
| Training Campaigns | (assignment only) | all | Multi-stage practice — launched from an assignment, not a menu |
| Scenario Builder | Training | trainer/admin | Author scenarios |
| Live Campaigns | Live Ops | trainer/admin | Real phishing simulations |
| Reply Review | Live Ops | trainer/admin | Approve AI replies |
Expand Down
Loading