wizz-method 2.21.0 → 2.22.0

package/src/skills-lib/wizz-vsl/references/diagnosis.md

@@ -0,0 +1,95 @@
+# Phase 2 · Market diagnosis (Schwartz)
+
+> Three questions, in this order, **before** any mechanism or lead. Getting this wrong means a technically correct VSL that speaks to a market that does not exist. When the Piva research exists, its awareness-levels report is the first evidence: quote it.
+
+## 1. What is the mass desire?
+
+Copy does not create desire. It **channels a desire that already exists** toward the product. If the desire is not in the market, no headline saves it.
+
+Pick **ONE** dominant desire, not three. Score each candidate from 1 to 5 on three axes and the highest sum wins (not the noblest, not the expert's favourite):
+
+| Axis | Question | Why it matters |
+|---|---|---|
+| **Urgency** | How strongly does she feel it *today*? | An urgent desire buys now |
+| **Permanence** | Felt for years, or a fad? | A permanent desire sustains an evergreen funnel |
+| **Reach** | How many in the audience share it? | Reach sets the scaling ceiling |
+
+**Tie:** highest urgency wins. If the audience has two sub-profiles with different desires, pick the dominant one for the lead and treat the other as a secondary layer inside the VSL. Never write one lead for both.
+
+**Test:** if a competitor could satisfy this desire tomorrow, would the audience buy from them? If yes, you found the right desire; the mechanism is what differentiates.
+
+## 2. Which awareness level is the market at?
+
+| Level | The audience… | What the LEAD has to do |
+|---|---|---|
+| **1. Unaware** | does not know it has the problem | Tell a story, dramatize a scene until they recognize themselves. No product. |
+| **2. Problem aware** | feels the pain, does not know a solution exists | **Name the pain with surgical precision.** The lead is the problem, not the promise. |
+| **3. Solution aware** | knows solutions exist, does not know yours | Promise the result and hint that a different path exists |
+| **4. Product aware** | knows your product, has not bought | Go to the mechanism or differentiator and break the real objection |
+| **5. Most aware** | knows and wants it, needs a push | Offer, price, deadline. Short copy. |
+
+**The VSL rule:** cold-traffic front end almost always runs at **2 or 3**. If the lead explains "why losing weight matters" to a level 3 audience, first-minute retention is gone, and the player chart will show it. Signal of a misreading: first-minute retention below 60%.
+
+## 3. Which sophistication stage is the market at?
+
+Sophistication measures **how many promises the market already heard**, not how much it knows.
+
+| Stage | State | What still works |
+|---|---|---|
+| **1** | Nobody promised anything yet | A direct, simple promise. Very rare today. |
+| **2** | Someone already promised | Enlarge the promise. |
+| **3** | Everyone promises the same thing | **New mechanism.** Not *what*, but *how*. This is where the VSL shines. |
+| **4** | Everyone has a mechanism | **Elaborate the mechanism:** easier, faster, surer, removing the old limitation. |
+| **5** | The market believes nothing | **Identification.** Sell who the person becomes. Proof and experience over promise. |
+
+**Awareness and sophistication are independent axes.** A market can know your product (awareness 4) and still be immune to any promise (sophistication 5). Read sophistication from Block 8: five competitors promising the same with named mechanisms means stage 4; "already heard it a thousand times and doesn't believe" means 5.
+
+## The table that governs the rest of the VSL
+
+| Awareness x Sophistication | What the VSL needs |
+|---|---|
+| 2 x 3 | Lead names the pain · problem mechanism is the heart · new, named solution mechanism |
+| 2 x 4-5 | Lead names the pain **from an angle nobody used** · elaborated mechanism + heavy early proof |
+| 3 x 3 | Lead promises result + hints at a different path · short problem mechanism, long solution |
+| 3 x 4 | Counterintuitive lead (contradicts the expected mechanism) · mechanism with the limitation removed |
+| 3-4 x 5 | Identification or story lead · social proof and demonstration replace the promise |
+| 4-5 x any | Short VSL, or no VSL: go to an offer page |
+
+**Short VSL** = the same four-block structure compressed, the pitch entering near the middle. When the low ticket ruler is in effect, its lengths replace these (SKILL.md, Step 0). State the chosen length in the script header.
+
+## Output · `01-diagnostico.md`
+
+```markdown
+# Diagnóstico de mercado
+
+## Desejo de massa dominante
+**<desejo em uma frase>**
+- Urgência: <alta/média/baixa> · porque <evidência do briefing>
+- Permanência: … | Alcance: …
+- Descartados: <outros candidatos e por que perderam>
+
+## Nível de consciência: <1-5>
+**Evidência:** <citação literal do briefing, da call ou da pesquisa do Piva>
+**Implicação para a lead:** …
+
+## Estágio de sofisticação: <1-5>
+**Evidência:** <o que o mercado já promete, segundo o Bloco 8>
+**Implicação para o mecanismo:** <novo / elaborado / identificação>
+
+## O que isso proíbe
+- <ex.: promessa direta de resultado está morta neste mercado>
+```
+
+**Honesty rule:** if you cannot cite evidence from the briefing for the level you assigned, the diagnosis is a guess. Say so out loud instead of faking confidence.
+
+## Gate 1
+
+Present the diagnosis with its evidence and ask explicitly:
+
+> "Confirma consciência **N** e sofisticação **M**? Se você acha que o mercado está em outro ponto, me diga: isso muda o mecanismo inteiro."
+
+Only after confirmation, record in `manifest.json`:
+
+```json
+{ "gate1_confirmado_em": "AAAA-MM-DD", "consciencia": 3, "sofisticacao": 4 }
+```

package/src/skills-lib/wizz-vsl/references/intake.md

@@ -0,0 +1,200 @@
+# Phases 0 and 1 · Ingestion and interview
+
+> The goal is to reach `00-briefing.md` with everything the VSL needs, nothing invented, every answer with its source. A good VSL is mostly briefing.
+
+## §0 · Ingestion: never ask what is already answered
+
+Before the first question, ask only this (in the user's language):
+
+> "Antes de te entrevistar: você já tem material sobre esse expert ou essa oferta? Transcrição de call, questionário preenchido, página de vendas antiga, VSL anterior, aulas gravadas, site, Instagram, documento de oferta? Me passa os caminhos ou os links que eu leio antes."
+
+Also check what the method already produced: an approved offer from `wizz-offer-forge` and the audience research from `wizz-pesquisa-de-publico-do-piva` (paths in SKILL.md, Step 0). They pre-fill whole blocks.
+
+When there is material:
+
+1. **Read all of it.** A call transcript is the richest source there is: the expert talking loosely about their own method beats any rushed questionnaire.
+2. **Pre-fill the briefing, marking the source of each answer:** `(call 01:07:22)`, `(questionario.md §12.1)`, `(OFFER-x.md §mecanismo)`, `(pesquisa-piva/voc.md)`.
+3. **Then interview only the gaps.** Nothing degrades a briefing faster than asking what the person already answered.
+4. **Return what you extracted before asking:** "do material eu já tirei X, Y e Z; me confirma e responde só o que falta".
+
+Without material, start at Block 1.
+
+## How to run the interview
+
+- **One block at a time.** Never dump sixty questions at once. Ask a block, receive, save to disk, move on.
+- **Numbered questions in the respondent's language,** without copy jargon ("qual seu USP?"). Ask like a curious friend.
+- **Offer audio.** "Se for mais fácil, grava um áudio respondendo, eu transcrevo." The expert talking loosely gives vocabulary, catchphrases and stories they would never write.
+- **"Não sei" is a valid answer.** It becomes `[LACUNA]` in the briefing, with one line on what that gap will cost the VSL.
+- **Dig into numbers and stories.** Every case: who, when, what changed, how much, and can we cite it publicly. Every number: its source.
+- **Save after each block** to `00-briefing.md`. An interview that lives only in the chat dies at the first context reset.
+- **Contradicting sources:** do not pick silently. Record both versions with their sources and take the divergence to the person. When in doubt, what the expert says **now** wins; the old material stays as a note.
+
+### Map: every answer has a destination
+
+| Block | Feeds |
+|---|---|
+| 1. Product and offer | the OFFER block (written first): modules, price, bonuses, guarantee |
+| 2. Audience | everything: who we speak to, in which words |
+| 3. Failed attempts | PROBLEM MECHANISM (the "what they tried" half) |
+| 4. Root cause | PROBLEM MECHANISM (the "why it could not work" half) |
+| 5. Method | SOLUTION MECHANISM (what it is + why it works) |
+| 6. Story | STORY block + credibility of the lead |
+| 7. Proof | proof type of the solution mechanism + `[DEPOIMENTO]` markers |
+| 8. Market and competition | sophistication stage (Phase 2) |
+| 9. Compliance | the LAW guardrail that crosses the whole VSL |
+| 10. Voice | vocabulary, rhythm, catchphrases |
+| 11. Operation | page rules, button delay, timer, upsell |
+
+## Block 1 · Product and offer `[MANDATORY]`
+
+*The offer is written first. Without this block closed, no line gets written.* When an approved `wizz-offer-forge` offer exists, this block is a confirmation, not an interview.
+
+1. What exactly does this VSL sell? Product name (or "no name yet").
+2. Format and access: recorded course, dated challenge, community, protocol, mentorship? Where and for how long?
+3. Modules, steps or deliverables, each with **what the person gains** from it, not only its name.
+4. Price. Is there an anchor: what would the equivalent cost in consulting, one-to-one or competition?
+5. Bonuses, and **where each came from** (the story of why it was created).
+6. Guarantee: period and conditions.
+7. A live workshop or event that can be offered as an urgency bonus?
+8. `[optional]` An immediate upsell after purchase? Which, and for how much?
+9. Who is this product **not** for?
+10. The triad: besides the content, does the buyer get some **recommendation** (what to do in their case) and some **follow-up** (support, group, AI, team)?
+
+## Block 2 · Audience `[MANDATORY]`
+
+*Pre-filled by the Piva research when it exists (VoC, psychographic profile, lexicon).*
+
+1. Who buys this? The real person: age, occupation, life moment, income.
+2. Who **pays you the most** today (real profile, not the desired one)?
+3. What is she feeling the day she watches this video? What happened that she is looking for this now?
+4. Which words does she use for the problem (hers, not the technical ones)?
+5. Beyond solving the problem, what does she **really** want? The dream behind the request.
+6. What keeps her awake? The biggest fear?
+
+## Block 3 · Failed attempts `[MANDATORY]`
+
+1. Everything this person **already tried** before finding you: products, methods, professionals, apps, courses, gurus.
+2. For each one: why it did not work (the technical explanation, not "she didn't try hard enough").
+3. How does she feel about failing so many times? Who does she blame?
+4. What has she heard so often she no longer listens? Which phrase in your market became a cliché?
+5. Why, with the problem knocking, does she keep putting the solution off?
+
+## Block 4 · Root cause `[MANDATORY]`
+
+1. In your expert opinion, what is the **real root cause** of the problem, the thing almost nobody sees?
+2. Which myth or "absolute truth" of your market would you take apart?
+3. The problem you solve is a consequence of what?
+4. Consequences of **not** solving it, on four fronts: physical, emotional, financial, relationships.
+5. Why does the problem keep growing despite so many solutions?
+
+## Block 5 · Method `[MANDATORY]`
+
+*The solution mechanism is born here.*
+
+1. Do you have your own method? Describe it.
+2. Its steps, in order. Do they have names? Do you see them as modules?
+3. If you had to summarize it in 3 to 5 pillars, which?
+4. **Why does your method work where the others failed?** The most important question of the interview: insist until the answer is specific, not "because it's personalized".
+5. Why has your customer never heard anyone put it this way?
+6. Is the method hard to execute? What does it demand?
+7. Time to the first result? To the full transformation?
+8. What happens to whoever follows it to the letter?
+9. Your promise in ONE sentence.
+
+## Block 6 · Story `[MANDATORY]`
+
+1. Did you **live** this problem, or **study or treat** people who lived it?
+2. If you lived it: life before, what you tried, how you found the way out.
+3. If you studied it: what you saw happening to so many people; a specific patient or student who changed your mind.
+4. Two or three cases of people you helped: before, what you did together, after.
+5. A failure of yours? (failure gives more credibility than success)
+
+## Block 7 · Proof `[MANDATORY]`
+
+1. How do you prove, to someone who doesn't know you, that it works?
+2. How many clients or students so far?
+3. Recorded testimonials, screenshots, reviews? Where?
+4. For each case in Block 6: before and after numbers? **Can we cite it publicly** (name, initials, city)? Is authorization needed?
+5. A study, survey or public data that supports your mechanism? Which, and where?
+
+> If 3 and 4 are "not yet", record it and mark `[DEPOIMENTO]` at those points of the VSL. Never write a fictional testimonial presented as real, not even "as an example". Missing proof does not block the rest of the script.
+
+## Block 8 · Market `[MANDATORY]`
+
+*This defines the sophistication stage in Phase 2. Pre-filled by the Piva trend report and angle matrix when they exist.*
+
+1. Competitors and references: what do they promise?
+2. What is **everyone** in your market saying right now? The standard promise?
+3. Which promise has the audience heard so often they no longer believe it?
+4. A stance in your market you refuse to adopt?
+5. `[optional]` If you ran ads or a VSL before: what worked, what died?
+
+## Block 9 · Compliance `[MANDATORY]`
+
+1. Anything you **cannot** promise or say publicly because of a professional council, regulator or legal rule (CFM, CFN, CRP, CFC, CVM, ANVISA, OAB)?
+2. Traffic platform restrictions for your niche (health, finance, before and after, income claims)?
+3. A forbidden term, number or kind of image?
+
+> The answers become a **GUARDRAIL (LEI)** header at the top of the VSL and override any copy rule of this skill.
+
+## Block 10 · Voice `[MANDATORY]`
+
+1. How would you describe your tone of voice? A catchphrase, an expression, a way of speaking that is yours?
+2. A word you **hate** and would never use?
+3. "Você" or "vocês"? Slang? Swearing?
+4. Two links of you speaking (Reels, YouTube, story) to calibrate rhythm.
+
+> With video or audio available, transcribe and extract a mini glossary of real vocabulary. Copy that sounds like the expert converts; copy that sounds like AI is recognized in three seconds.
+
+## Block 11 · Operation `[optional]`
+
+1. Video player (VTurb, Panda, unlisted YouTube)? Can the button delay be set?
+2. Where is the page hosted? Checkout (Hotmart, Kiwify, Eduzz, Stripe)?
+3. One-click upsell? Order bump?
+4. A timer available for the urgency bonus?
+
+## Phase 1 exit
+
+Before Phase 2, show the human this summary:
+
+```
+BRIEFING · completude
+✅ Blocos completos: 1, 2, 3, 4, 5, 6, 9, 10, 11
+⚠️ LACUNAS que impactam a VSL:
+   - [LACUNA 7.4] Sem autorização de uso dos casos → a VSL fica com [DEPOIMENTO]
+     em 3 pontos e a prova social do pitch fica fraca. Custo: conversão do pitch.
+   - [LACUNA 8.3] Não sabemos que promessa o mercado já não acredita → a
+     sofisticação vai ser hipótese. Custo: risco de mecanismo gasto.
+▶️ Sigo pra Fase 2 assim, ou você fecha essas lacunas antes?
+```
+
+A gap in a mandatory block does not stop the work, but it must not stay silent: say what is missing and what it costs. The human decides to go on with the gap, informed.
+
+## Template · `00-briefing.md`
+
+```markdown
+# Briefing · <expert> / <produto>
+
+> Fontes: <arquivos lidos, oferta aprovada, pesquisa do Piva>. Entrevista em <data>.
+> Toda resposta traz a fonte entre parênteses. `[LACUNA]` = não respondido.
+
+## 1. Produto e oferta
+- **Nome:** … (call 00:12:40)
+- **Preço:** R$ … | **Âncora:** …
+- **Módulos:** …
+- **Bônus:** … (cada um com a história de origem)
+- **Garantia:** …
+- **Tríade:** conhecimento … | recomendação … | acompanhamento …
+
+## 2. Público
+## 3. Tentativas fracassadas
+## 4. Causa raiz
+## 5. Método
+## 6. História
+## 7. Provas
+## 8. Mercado
+## 9. GUARDRAIL (LEI) · proibições
+## 10. Voz · vocabulário real
+## 11. Operação
+## Lacunas em aberto
+```

package/src/skills-lib/wizz-vsl/references/mechanisms.md

@@ -0,0 +1,138 @@
+# Phase 3 · Problem mechanism and solution mechanism
+
+> **The whole VSL derives from the mechanism.** A mistake here costs re-recording the video, so this phase ends in a decision document and a human gate, never in your own choice. When an approved `wizz-offer-forge` offer already names the unique mechanism, it enters as the master solution mechanism by default (marked ⭐ with its source); the 5x5 still runs on the problem side and offers alternatives on the solution side.
+
+## What each one is
+
+**PROBLEM mechanism:** the explanation of *why this person failed until today*, and why it is not her fault. It externalizes the blame **truthfully**, names the common enemy and turns accumulated frustration into attention.
+
+> Formula: **what she already tried + why it could not work = the root cause**
+
+**SOLUTION mechanism:** the explanation of *why the expert's method solves what the others did not*. Nameable, specific, defensible.
+
+> Formula: **WHAT the method is + WHY it works.** The **HOW** stays for the offer.
+
+Keeping the "how" back is what creates the need to buy. A VSL that teaches the step by step gives the value away and kills the pitch.
+
+### What a mechanism is NOT
+
+| Not a mechanism | Why |
+|---|---|
+| "My method is personalized" | Everyone says it. Spent mechanism (sophistication 4+). |
+| "You need consistency or discipline" | Hands the blame back to the viewer. |
+| "A simple 5-step process" | That is product structure, not mechanism. |
+| "Because I have 15 years of experience" | That is a credential. A credential does not explain a cause. |
+| A pretty metaphor with no cause inside | A metaphor is packaging. |
+
+## Traceability rule
+
+When the source material (call, questionnaire, class, approved offer) **already contains** an articulated mechanism, it is the **master mechanism**: extract it and cite it with timestamp or line. Your synthesis never replaces it; your original proposal enters marked `🆕 proposta nova (não veio da fonte)` and the human decides. The source mechanism enters as master by default and as your recommendation; the human may prefer another, as an explicit decision of theirs, never a silent substitution of yours.
+
+## The 5x5 format
+
+Produce **5 problem mechanism proposals** and **5 solution mechanism proposals**. Five forces you out of the obvious: the first one is always the most predictable. Each proposal has exactly these fields:
+
+```markdown
+## P3 · <nome curto e memorável> ⭐/🆕
+**Articulação:** <3 a 6 linhas, escritas como seriam ditas na VSL, nunca descrição em terceira pessoa>
+**Hook exemplo:** *"<uma frase que abriria a VSL com esse mecanismo>"*
+**Rastreabilidade:** ✅ <fonte + timestamp> | 🆕 proposta nova
+**Schwartz:** <por que serve pra consciência N e sofisticação M>
+**Risco:** <o que precisa ser verdade pra funcionar, ou onde quebra>
+```
+
+⭐ = came articulated from the source · 🆕 = your new proposal.
+
+### Angles for the 5 problem proposals
+
+1. **The math that does not add up:** she is rational, the cost is today and the reward is abstract.
+2. **The wrong ruler:** she got results and did not see them, because she measured with the slowest indicator.
+3. **Someone else's plan:** everything she received was designed for another profile.
+4. **The invisible part:** the market sells the visible slice of the problem and ignores the rest.
+5. **The inverted order:** she started at step 3; steps 1 and 2 never happened.
+6. **The system against her:** something structural (industry, algorithm, law, biology) works against her result and nobody told her.
+
+### Angles for the 5 solution proposals
+
+The mechanism is the **variable only this method controls**:
+
+1. **The right order:** the same elements, in the sequence nobody respects.
+2. **The ignored variable:** what everyone treats as detail is the lever.
+3. **The measurement:** the method gives a new ruler that shows progress before the obvious result.
+4. **The old limitation removed:** what made it unfeasible (time, schedule, cost, discipline) no longer exists, and the mechanism explains how.
+5. **Diagnosis before prescription:** the method first finds out how *this person* works.
+6. **The layer nobody delivers:** the market sells the visible part; the method works the invisible one.
+
+Each one answers "why it works" with **one of three proof types**, and the type follows the niche, not the expert's taste ([niches.md](niches.md)):
+
+| Proof type | When | Format |
+|---|---|---|
+| **Scientific** | Health, performance, aesthetics, nutrition | Study, physiological mechanism, exam |
+| **Logical** | Finance, business, legal, B2B | Math, cause and effect, side-by-side simulation |
+| **Social** | Life change, hobby, relationship, volume | Cases, screenshots, number of students |
+
+If the expert **lacks** the proof the niche asks for, that is a finding: record it as the proposal's risk, pick the mechanism that stands on the proof that exists, and tell the human what would have to be produced for the ideal mechanism.
+
+**What/why versus how boundary:** a sentence that answers *why it works* belongs in the VSL; one that answers *what to do on Monday morning* belongs in the offer. Test: can the viewer execute it alone after hearing it? If yes, you gave away the how: cut it.
+
+### Proper name
+
+Every approved solution mechanism gets a name: "Protocolo X", "Método das 3 Réguas". A name turns the mechanism into property, repeatable in the lead, the ad, the email and the page. A good name names the *cause* or the *turn*, not the category.
+
+## Pairing: not every P fits every S
+
+Before recommending, test the pair:
+
+- Does the S solve **exactly** the cause the P named? (if P says "you measured with the wrong ruler", S has to deliver a new ruler, not "personalized follow-up")
+- Does the pair survive the diagnosed sophistication?
+- Does the pair fit the speaking time of the problem and solution blocks under the ruler in effect?
+- Does it respect the compliance guardrail of Block 9?
+
+Recommend **one pair**, with one sentence on why this pair and not the others. Offer one or two secondary layers (reinforcement, not master).
+
+## Adversarial stress test (maker is not checker)
+
+Before presenting, hand the draft to a **different** agent, or, failing that, run these six questions yourself with deliberate ill will:
+
+1. Has the market heard this? (if yes, it is spent: kill it)
+2. Is it falsifiable, or vague enough to never be wrong? (vague = weak)
+3. Do I overpromise? Does a sentence become a legal or compliance problem?
+4. Is the master mechanism traceable to the source, with timestamp?
+5. If I swap the expert's name, does it still fit? (if yes, too generic)
+6. Can a skeptical viewer rebut it in one sentence? Which one? Does the VSL answer it?
+
+## Gate 2
+
+Present `02-mecanismos-5x5.md` as a **decision document**:
+
+> "10 propostas. Preciso que você escolha **1 mecanismo-mestre do problema** e **1 da solução**. Minha recomendação: **P1 + S3**, porque … As demais estão numeradas pra você trocar. Camadas secundárias sugeridas: P2."
+
+Only after the explicit choice, record:
+
+```json
+{ "gate2_aprovado_em": "AAAA-MM-DD", "aprovado_por": "<nome>",
+  "mecanismo_problema": "P1", "mecanismo_solucao": "S3", "camadas": ["P2"] }
+```
+
+**Phase 4 reads this field and refuses to run while it is empty.** A draft becomes a recording.
+
+## Template · `02-mecanismos-5x5.md`
+
+```markdown
+# Mecanismos 5×5 · <expert> (documento de decisão)
+
+> Diagnóstico que governa as 10 propostas: consciência N, sofisticação M, desejo dominante <…>.
+> Gate: escolher 1 mestre de cada lado → registrar no manifest → só então escrever a VSL.
+
+## Diagnóstico resumido
+# PARTE 1 · MECANISMO DO PROBLEMA
+## P1 … P5
+# PARTE 2 · MECANISMO DA SOLUÇÃO
+## S1 … S5
+# RECOMENDAÇÃO
+**Par recomendado:** P_ + S_ · porque …
+**Camadas secundárias:** …
+**Pares que eu descartaria:** … porque …
+# STRESS-TEST
+<as 6 perguntas respondidas contra o par recomendado>
+```

package/src/skills-lib/wizz-vsl/references/niches.md

@@ -0,0 +1,40 @@
+# Adapting to the niche
+
+> Classify the niche in Phase 1 and apply it in every phase. The same framework written in the wrong tone converts half.
+
+| Niche | Lead tone | Problem mechanism | Proof type | Typical front-end price |
+|---|---|---|---|---|
+| **Finance / investing** | Logical, direct, little emotion. Negative leads dominate | Fast and factual (broken system, inflation, math) | **Logical** | R$ 197-300 |
+| **Health / fitness / aesthetics** | Emotional, deep, validates the frustration | Slow, digs the root cause, "não é culpa sua" | **Scientific** | R$ 97-197 |
+| **Life change / digital business** | Aspirational with proof | Opportunity cost, lost time | **Social** | R$ 97-197 |
+| **Professional (legal, medical, accounting)** | Sober, technical, no hype | Outdated method, career risk | **Logical + authority** | R$ 297-589 |
+| **Volume / hobby / low ticket** | Light, close, community | Simple and short | **Social** | R$ 40-67 |
+
+Price ranges are calibration, not copy. When the low ticket ruler applies, its price and guarantee ruler (through `wizz-offer-forge` and `wizz-lowticket-metodologia`) replaces this column.
+
+## Notes per niche
+
+**Finance.** Broken-institution and math-comparison leads are the most resilient. Return promises are regulated in many countries.
+
+**Health.** The problem block is the longest of the VSL, not the shortest. The audience arrives defeated by previous attempts; validating the frustration *before* proposing anything buys the attention. Compliance is risk number one.
+
+**Life change.** High sophistication, a market saturated with promises. Usually asks for an elaborated mechanism (stage 4) or identification (5).
+
+**Professional.** The buyer has a reputation at stake. Hype kills conversion. Authority, technical specificity and cases with numbers do the work.
+
+**Volume.** Do not force a high-ticket structure. Short VSL, simple offer, focus on recurrence and community. The game is scale, not margin. When the operation is low ticket, the ruler in SKILL.md, Step 0, decides the runtime.
+
+## Compliance: the guardrail that beats every copy rule
+
+Block 9 becomes a **LEI** header at the top of the VSL. No technique of this skill authorizes breaking it.
+
+| Area | Typical restriction |
+|---|---|
+| Health, nutrition, psychology, dentistry | Councils often forbid promising results, before and after, guaranteed cure, patient images |
+| Finance and investing | Return promises, personalized advice without registration |
+| Traffic platforms | Meta and Google restrict income, health and weight-loss claims, and accusatory "você" |
+| Third-party data | A student's or patient's name, image and numbers need recorded authorization |
+
+Safe patterns under restriction: third-person scenes instead of direct accusation · a statistic with a cited source instead of diagnosing the viewer · an experience account instead of a result promise.
+
+When the expert cannot answer Block 9, do not invent the answer: record `[LACUNA]` and write in **conservative mode**: no result promise ("você vai emagrecer X kg" becomes "o que o método faz"), no second-person diagnosis (becomes a third-person scene), no before and after, no number without a cited source, no third-party name. State in writing that the VSL is in this mode and what can be released once someone confirms the niche's rules.

package/src/skills-lib/wizz-vsl/references/offer.md

@@ -0,0 +1,91 @@
+# Phase 4.1 · The OFFER (written first)
+
+> Counterintuitive and non-negotiable: **the offer is the first block written**, even though it is the last the viewer sees. Prerequisite: `manifest.gate2_aprovado_em` filled in. Check it before writing.
+>
+> When an approved `wizz-offer-forge` offer exists, it is the source for steps 2 to 13 (name, stack, bonuses, guarantee, anchor): this block turns it into **spoken** script, it does not redesign it. A change of price, bonus or guarantee goes back to `wizz-offer-forge` and reopens Gate 2.
+
+Default position: the pitch enters at the 12 to 14 minute turn and runs 4 to 6 more (full VSL 16 to 20). Under the low ticket ruler, the pitch enters at about two thirds of the runtime the ruler sets.
+
+## The 15 steps, in order
+
+**1. Transition from the mechanism to the offer.**
+> "Agora que você entendeu <mecanismo da solução>, deixa eu te mostrar como aplicar isso na sua vida a partir de hoje."
+
+Never cut dry to "so I created a course".
+
+**2. Product name.** Strong and specific. Words that convey a system (*protocolo, método, sistema, programa, desafio*). Avoid *curso* and *formação*: they sound like content, not result.
+
+**3. Who it is for, and who it is not for.**
+> "Isso é pra você se <condição 1>, <condição 2> e <condição 3>. E não é pra você se <condição>."
+
+The negative qualification raises perceived value and buyer quality (Block 1.9).
+
+**4. What the person receives.** Module by module, always **benefit, not feature**:
+> ❌ "Módulo 3: Alimentação"
+> ✅ "Módulo 3: o que comer nos dias em que a agenda vira do avesso, pra você não recomeçar do zero toda segunda"
+
+**5. The triad.** Strong products communicate three layers:
+
+| Layer | Line | Example |
+|---|---|---|
+| **Knowledge** | "Eu te ensino a pescar" | the content, straight to the point |
+| **Recommendation** | "Eu te dou o peixe" | what to do in her specific case (diagnosis, plan, AI) |
+| **Follow-up** | "Eu te acompanho na pescaria" | continuous support: group, WhatsApp, AI, team |
+
+Name the layers that exist and never invent the missing ones. If two are missing, flag it to the human: it is an offer weakness, not a copy one.
+
+**6. Interleaved testimonials.** `[DEPOIMENTO 1]`, `[DEPOIMENTO 2]`… spread **between** modules, never piled in one block. When there is no proof material, the markers hold the place, or the block is omitted; the rest of the pitch goes on.
+
+**7. Price anchoring.** Before revealing the price, compare with the real, expensive alternative:
+> "Uma consultoria individual comigo sai R$ X por sessão. Um acompanhamento de 3 meses passa de R$ Y."
+
+Anchor only against a real, supported alternative. A charging history or discount that never existed is never stated as fact.
+
+**8. Price with contrast** · `[BOTÃO APARECE AQUI]`
+> "Mas hoje não é nada disso. É R$ <preço>."
+
+The button appears **here**, when the price is revealed: that is what the player delay waits for.
+
+**9. Bonus with a 60-second timer.** A real urgency bonus with a visual countdown, typically a live workshop seat. It is one of the strongest levers of the format. Honesty requirement: the bonus exists and is delivered. A timer that resets and keeps offering destroys trust in the whole funnel.
+
+**10. Guarantee.** Period and condition, in one clear sentence that matches the real terms exactly (an encouraged usage challenge is not automatically a refund condition).
+
+**11. Extra bonuses, with a story.** Each bonus gets 2 or 3 lines of origin. A bonus listed without story is noise; with story it is value.
+
+**12. Express stack.** Quick recap, in a list, with value per item:
+> "Programa completo: R$ X · Bônus 1: R$ Y · Bônus 2: R$ Z · Total: R$ <soma>. Hoje: R$ <preço>."
+
+**13. Surprise gift.**
+> "E tem uma coisa que eu nem coloquei na lista: assim que você entrar, recebe um convite pra <aula/encontro>."
+
+It raises the open rate of the first post-purchase message, where the whole backend starts.
+
+**14. Emotional close.** Go back to the lead's problem and project the two futures.
+
+**15. Final CTA.** One instruction, no ambiguity: "Clica no botão aqui embaixo agora." `[CTA → aponta pro botão que já apareceu no passo 8]`
+
+## The value equation (apply to every line of the pitch)
+
+Every element of the offer pushes at least one of: **↑ dream outcome · ↑ perceived likelihood · ↓ time to result · ↓ effort and sacrifice**. A bonus or module that moves none of the four weakens the offer by dilution.
+
+## Page and player rules (they move conversion more than copy)
+
+These belong to the page around the player, written by `wizz-copywriting`; list them in the hand-off.
+
+| Rule | Why |
+|---|---|
+| **No headline above the video** | Removing the headline improves play rate |
+| **No email or phone form before the VSL** | Capturing first lowers front-end conversion |
+| **Delayed button**, only after the price | An early button anticipates the pitch and lowers retention |
+| **No discount button or coupon on the page** | Discount buyers take the upsell far less |
+| Traffic goes straight to the open page | No intermediate capture step |
+
+Under the low ticket ruler, its page and player shards (`06b-player-vsl.md`, `12c-anatomia-low-ticket.md`) win over this table.
+
+## Output of this step
+
+The offer block in **spoken language** (it is a script, not page text), with `[DEPOIMENTO]`, `[BOTÃO APARECE AQUI]` and `[A DEFINIR]` markers where information is missing.
+
+## OTO and downsell videos
+
+A short upsell video after purchase follows a different sequence, in [blocks-and-leads.md](blocks-and-leads.md) §OTO. It never implies that the buyer accepted another optional offer.

package/src/skills-lib/wizz-vsl/references/production.md

@@ -0,0 +1,75 @@
+# Phase 6 · Production plan and hand-off
+
+> Animation makes a clear script easier to follow; it cannot rescue vague writing. Production starts only after the script-only audit and the claims ledger are closed ([assembly-qa.md](assembly-qa.md)). This skill **plans** the production and hands it off: drawing, voicing and assembling the video belong to the designer area. The copy agent never invokes another area's agent: offer `wizz-maestro` when the user wants the video in the same delivery.
+
+## 1. Beat sheet · `05-beat-sheet.md`
+
+One row per beat, in presentation order. It is the contract the production reads.
+
+| # | Narration (exact) | On-screen line | Visual evidence | Claim supported | Source |
+|---|---|---|---|---|---|
+| 1 | "…" | ≤ 6 words | product crop / real screen / `[B-ROLL: …]` | ledger #3 | `00-briefing.md` 7.4 |
+
+- Capture product states only after confirming the script and the UI agree. Label fixture or example data honestly. Never expose customer data, credentials or live keys.
+- A small visual system: two type roles, a restrained palette, consistent safe margins, a few repeatable scene types (intro, problem, reveal, demo, proof, guarantee, offer, CTA). Patterns, not required branding: the project's brand wins.
+- No durations by hand: every beat's length comes from the final audio (section 3).
+
+## 2. Narration
+
+- Record or generate the narration from the **exact** final script. Paid voice generation only with the user's explicit authorization; keep the complete inputs (text, voice and model settings, request IDs, returned alignment, audio hashes) so an uncertain job is resumed instead of paid twice.
+- When revising, regenerate only the changed sentences if the voice splices cleanly, with adjacent text as prosody context. Listen to every seam; replace the whole paragraph only when the splice stays audible.
+
+## 3. Timing from the final audio
+
+Invoke the `wizz-ctc-align` skill on the final narration and the exact spoken text: word-by-word timestamps snapped to real silences become the single timing source. Never estimate timing by hand and never use Whisper timestamps for a known script.
+
+- Scene durations are built from that alignment.
+- Big on-screen lines change on **spoken beats**, anchored to the word timings, never to arbitrary seconds.
+- If a section is cut after alignment, update audio, word timings, captions, scene hosts, animation anchors, music ducking and the final fade **together**. Old offsets become invalid after an upstream edit.
+
+## 4. Drawing and assembly (designer area)
+
+The method's video pipeline rule: **HyperFrames draws the screens (silent visual), Remotion assembles the film (audio, captions, frame timing).**
+
+- `hyperframes` draws each silent scene from the beat sheet: kinetic headline cards, animated stack, price reveal, guarantee card, real product mockups.
+- The `wizz-remotion-best-practices` skill assembles the scenes with the narration, captions, SFX and the motion anchored to the `wizz-ctc-align` timings.
+- Effects and sound by semantic event, if wanted: the `wizz-video-effects` skill, after the assembly plan.
+
+### Motion rules for the big lines
+
+- Split on-screen headlines into one to three short lines, each a complete, concrete thought.
+- Enter lines in sequence with fast, damped motion; use scale, opacity or horizontal movement consistently.
+- Highlight one decisive line or word with the accent colour; never colour every benefit.
+- Keep copy readable after its entrance. Motion marks a beat, reveals evidence or guides focus.
+- During demos the UI dominates: crops and pans focus attention without hiding the state needed to verify the claim. When the finished output is the strongest proof, play it unobstructed and long enough to judge.
+- Respect title-safe areas, captions, mobile scaling and reduced motion.
+
+**Illustrated VSL (optional reference).** For a slide-by-slide stickman or illustrated VSL with a timed image manifest and frame verification, `mikefilsaime-groove/stickman-vsl-director` (MIT) is a reference for the storyboard grammar (one image per narration span, a mandatory art-direction choice, a review deck before rendering). It is not installed by the method; its own scripts are not run from here.
+
+## 5. Verification of the rendered master
+
+- Inspect frames from the rendered master, not only development snapshots.
+- Watch with sound from start to end; listen at every splice and transition.
+- Check captions against the final audio and the visible claims.
+- No black gaps, clipped lines, stale frames, drift, missing fonts or unreadable demo crops.
+- Measure streams, dimensions, runtime, loudness, peaks, file size and hashes when relevant. Keep a high-quality master separate from the web encode.
+
+## 6. Player QA (on the real page)
+
+Test the actual page player on phone width and desktop: explicit sound start, restart behaviour, native controls, captions, close or escape, pause when out of viewport, poster, reduced motion, and the button delay set to the price moment (`[BOTÃO APARECE AQUI]`). Under the low ticket ruler, `06b-player-vsl.md` of `wizz-lowticket-metodologia` sets what the player must deliver.
+
+Report the measured runtime and the remaining limitations. Nothing is published, deployed, or sent to paid generation without the user's authorization.
+
+## Hand-off block
+
+Close Phase 6 with this block in the reply:
+
+```
+🎬 HANDOFF DE PRODUÇÃO · <produto>
+Roteiro final: {output_folder}/vsl/<slug>/03-vsl-principal.md (lead default: <qual>)
+Beat sheet: {output_folder}/vsl/<slug>/05-beat-sheet.md (<N> beats)
+Ledger de claims: 04-qa.md (<N> claims, <N> com marcador pendente)
+Régua: <default | lowticket: duração e posição do pitch>
+Ordem: narração final → wizz-ctc-align → hyperframes (cenas mudas) → Remotion (montagem) → QA do player
+Página em volta do player: wizz-copywriting com as regras de página e player
+```