moveros 4.0.8 → 4.1.0

package/src/workflows/ignite.md

@@ -1,720 +1,720 @@
----
-description: Initiates a new project or feature. Full pipeline from interrogation to Brief + Plan in one workflow.
----
-
-# Ignition Protocol (V4)
-
-**Trigger:** User runs `/ignite`, `/ignite --fast`, `/ignite --feature`, or `/ignite --monetize`
-**Role:** Systems Architect (single adaptive persona).
-**Task:** Transform a vague idea into a locked project with Brief, Plan, and Repo. Or add a feature to an existing project.
-**Risk Tier:** MEDIUM
-
----
-
-## 0. PRE-FLIGHT
-
-Global Rules are pre-loaded as CLAUDE.md / GEMINI.md. Output `[Rules Loaded - DATE]`.
-
-LOAD (Memory Hierarchy):
-1. Auto_Learnings.md (P0)
-2. Active_Context.md (P1)
-3. Strategy.md (P2)
-4. Identity_Prime.md (P3)
-5. Mover_Dossier.md (P3)
-6. Last 3 Daily Notes (P5) — read via subagent (model: haiku). Needed for PIVOT DETECTION (Step 0.5). Shows what the user has actually been working on. If the new project contradicts their current Strategy or today's plan, this is the evidence.
-
-CONSTRAINTS:
-- All recommendations grounded in user's Strategy and Dossier
-- Check Dossier before suggesting cold starts [Mover_Global_Rules.md: S1]
-- Never proceed to next step without explicit user approval
-- Be proactive and strategic, not a checklist follower
-- Windows: no && in commands
-
----
-
-## 1. MODE DETECTION
-
-**Fast Mode:** If `/ignite --fast`: FAST MODE (no TDD, hard-coded values OK, one file OK).
-If keywords: "quick script", "throwaway", "prototype" -> Suggest Fast Mode.
-
-**Feature Mode:** If `/ignite --feature` OR user describes a feature for an EXISTING project:
-1. Detect the target project. Check for existing `project_brief.md` and `plan.md` in the project folder.
-2. If no existing project found: "No project context found. Run `/ignite` for a full project, or specify which project folder."
-3. LOAD existing `project_brief.md`, `plan.md`, `project_state.md` — these define the system the feature lives in.
-4. Skip Clusters 4 (Business Model) and 5 (Resources) — already established for the project.
-5. Adapt remaining clusters to focus on the feature within the existing system:
-   - Cluster 1 → Feature Vision (what, why, how it fits the existing product)
-   - Cluster 2 → User impact (who benefits, what changes for existing users)
-   - Cluster 3 → Technical integration (what touches, what breaks, blast radius)
-   - Cluster 6 → Edge cases specific to this feature
-   - Cluster 7 → Stress test against existing system constraints
-6. Output: Feature spec appended to `project_brief.md` under `## Feature: [Name]` + new Phase added to `plan.md`.
-7. Skip Git Init (repo already exists).
-
-**Monetize Mode:** If `/ignite --monetize` OR user mentions revenue, selling, customers, pricing, "make money from this":
-1. This mode fires Phase 0 (Market Validation) BEFORE the normal interrogation clusters.
-2. Phase 0 spawns research subagents in parallel to build market intelligence.
-3. A validation gate blocks proceeding to Cluster 1 until the user reviews market findings.
-4. After Phase 0 approval, the normal interrogation proceeds with market context loaded.
-5. Clusters 2 (User & Market) and 4 (Business Model) are pre-populated from Phase 0 research — the AI uses research findings as the baseline and the user corrects/extends, rather than starting from zero.
-
-**New Project Mode (default):** If neither flag, ask: "Is this a prototype or production code? If you intend to sell this, consider running `/ignite --monetize` instead."
-
----
-
-## 1.5 PIVOT DETECTION (All modes except --feature on existing project)
-
-Before proceeding, cross-reference the new project against the user's current state:
-
-1. **Read today's Daily Note** (if exists): What is the plan? What is the Single Test?
-2. **Read last 3 Daily Notes**: What has the user been working on? Is there an active project with momentum?
-3. **Read Strategy.md**: What is the current hypothesis being tested?
-4. **Read Active_Context.md**: What is the stated bottleneck?
-
-**If the new project contradicts any of these:**
-
-Apply Escalating Friction (Global Rules S1.5):
-
-- Level 1: "Your Strategy says [X]. Your plan today says [Y]. You're proposing a new project: [Z]. Is this a deliberate pivot or are you avoiding [Y]?"
-- Level 2 (if they push through): "What about [current project] makes you want to start something new? One sentence."
-- Level 3 (if they still push through): "Noted. Logging this as a potential pivot. We'll proceed, but `/analyse-day` will flag this."
-
-**If not contradictory** (e.g., the new project IS the strategy, or no active projects exist): proceed without friction.
-
-**Log override if applicable:** Write `[OVERRIDE] User chose /ignite [new project] over [plan item]. Reason: "[user's stated reason]"` to today's Daily Note Session Log.
-
----
-
-## 1.7 PHASE 0: MARKET VALIDATION (--monetize only)
-
-> **Gate:** Do not write a single line of code, generate a brief, or start architecture until Phase 0 is reviewed and approved. Building without validation is the most expensive form of procrastination.
-
-### Step 1: Spawn Research Subagents (Parallel)
-
-Fire 3 research subagents simultaneously using the Task tool. All agents should use `model: "sonnet"`:
-
-**Agent A — ICP & Pain Points** (skill: `agent-research-analyst`)
-- Search Reddit (r/[relevant subreddits]), Hacker News, relevant forums for pain points
-- Collect verbatim quotes (exact language the target market uses — this becomes copy later)
-- Identify emotional tone per pain point (frustration, shame, resignation, hope)
-- Rank pain points by signal strength (upvotes, comment depth, recurrence)
-- Estimate willingness to pay based on existing spend patterns
-
-**Agent B — Competitor & Market Analysis** (skill: `agent-research-analyst`)
-- Map direct competitors (same niche + same solution type)
-- Map adjacent competitors (same niche, different solution OR different niche, same solution)
-- For each competitor: what they charge, how many users, what's missing (the gap)
-- Identify white space — what nobody is building
-- Market size estimate (TAM, SAM, SOM)
-
-**Agent C — Go-To-Market & First Users** (skill: `agent-content-researcher`)
-- Where do the first 50 customers come from? (specific channels, not "social media")
-- What distribution channels exist? (communities, newsletters, directories, marketplaces)
-- What launch playbooks have worked for similar products? (Product Hunt, Reddit, X threads)
-- What's the cheapest way to validate demand before building?
-- Map the Micro-Action-Chain (every step from first exposure to purchase). Identify stimulus fatigue risks and contrarian positioning opportunities
-
-**Agent D — Messaging & Psychology** (skills: `marketing-psychology`, `copywriting`, `pricing-strategy`)
-- Take the pain points from Agent A and translate into messaging:
-  - Primary message (1 sentence that captures the core promise)
-  - 3 supporting messages (each addressing a top pain point)
-  - Objection handling (top 5 reasons someone won't buy + the counter)
-- Apply psychological leverage analysis:
-  - Identify the top 10 cognitive biases most relevant to this niche and product (from the 88 in the leverage list)
-  - Apply Lock & Key Theory: the human mind is the lock, the stimulus is the key. Build the stimulus to fit the lock, not the other way around.
-  - Check for Lollapalooza Effect opportunities: where can 3+ biases stack to create non-linear persuasion?
-  - Flag any overdone-it risk (too many levers = slimy, not genuine)
-- Pricing analysis: what price anchors exist in this market? What's the perceived value vs cost?
-- Positioning: "We are the only [X] that [Y]" — find the only statement
-- Tone/voice: based on ICP, what voice resonates? (technical, casual, authoritative, peer-to-peer)
-
-### Step 2: Market Validation Frameworks
-
-While research agents run, apply these frameworks to the user's idea:
-
-**Niche Selection Checklist**:
-
-Ask the user these keystone questions (all must pass or the niche is unviable):
-1. Does the niche need what you are selling? (Pain drives acquisition)
-2. Is this niche growing? (Market must be expanding)
-3. Can this niche afford your price? (Look at competitor pricing as signal)
-4. Are you interested in this niche? (You'll spend thousands of hours here)
-5. Are there more than 30,000 people in this niche?
-6. Will you have competitors already succeeding? (Competition = sign of life)
-7. Can people in this niche be easily found and contacted?
-
-**STOP. Wait for user response.**
-
-If ANY keystone question fails: "This niche fails the viability check on [X]. Either pivot the niche or acknowledge the risk before proceeding."
-
-**Client Avatar Worksheet**:
-
-Go deeper than demographics. Ask the user (or infer from Agent A research):
-1. **Meet Your Client:** What do they do day-to-day? How do they spend their week? Where do they get their news/information?
-2. **Financial & Emotional State:** How much do they make? How do they feel about their work? What are their concerns in life?
-3. **History:** What happened in the past that led them here? A former job, a failed attempt, a frustrating experience?
-4. **Decision Trigger:** What thoughts go through their head right before they buy? What's the "final straw" that makes them pull the trigger?
-5. **Product Fit:** What about YOUR product matters to THIS person? How does it solve their pain or improve their life?
-
-**STOP. Wait for user response.**
-
-**Offer Creation Worksheet**:
-
-Walk the user through:
-1. **Define Who:** Which specific segment of the niche? What's their current situation vs desired situation?
-2. **Define Outcome:** What's the bold promise? Make it tangible (numbers-based).
-3. **Define Timeframe:** How long to deliver the outcome? Can it be shorter?
-4. **Methodology:** What steps/components deliver the outcome? Simplify to 6-9 words each.
-5. **Factors of Value:** For each methodology step, list obstacles the prospect will imagine. Invert each obstacle into a solution. This is how value is created.
-6. **Risk Reversal:** What guarantee can you offer? The bolder the guarantee, the higher the conversion.
-7. **Packaging:** Reposition from "what it is" to "what it means." Not "markdown files" — "a self-improving execution engine."
-
-**STOP. Wait for user response.**
-
-**Offer Asset Creation**:
-
-Using the offer framework above + pain points from Agent A, create:
-
-1. **Short-Form Messages (15-20):** 3-6 word hooks. Focus on end result. Reverse-engineer from the latent conditions of the market. Write what they want to read, not what you want to write. Examples: "escape the matrix", "make your bed", "become an acquisition expert."
-2. **Long-Form Messages (4-5):** 1-4 sentence summaries of the offer. Include: who you help, end result, timeframe, risk reversal. Use words that display extreme confidence. Don't discuss the 'how.'
-3. **Strategy Document:** List methodology items + factors of value. Less is more. This becomes the terms of your guarantee and sets deliverable expectations.
-4. **The Pitch (5-part):** Our Promise, Our Strategy, Our Guarantee, The Ultimatum (either/or), The Packaging ("think of us like..."). Must be readable in under 2 minutes. Does NOT include the price.
-
-**STOP. Wait for user response.**
-
-### Step 3: Compile Market Intelligence
-
-Once research agents return, compile into `dev/marketing/` (or project-specific `dev/marketing/` if a project folder exists):
-
-| File | Contents |
-|------|----------|
-| `ICP.md` | Ideal Customer Profile (3 tiers: primary, secondary, tertiary). Anti-ICP. Pain points ranked. Willingness to pay. |
-| `client_avatar.md` | Deep avatar profile: demographics, psychographics, history, decision triggers, "final straw" moment, product fit. |
-| `competitor_map.md` | Direct and adjacent competitors. Gap analysis. White space. Key people. |
-| `pain_points_raw.md` | Verbatim quotes organized by pain category. Emotional tone map. Sources. |
-| `gtm_plan.md` | First 50 customers strategy. Channels. Launch playbook. Validation experiments. |
-| `messaging_framework.md` | Primary message. Supporting messages. Objection handling. Psychology levers. Voice/tone. Positioning statement. |
-| `offer_framework.md` | Offer worksheet completed. Niche viability score. Packaging. Guarantee. Pricing. |
-| `offer_assets.md` | 15-20 short-form messages, 4-5 long-form messages, strategy document, 5-part pitch. |
-| `acquisition_system.md` | Micro-action-chain. Contrarian attraction analysis. System design (7-step). Sub-system priorities. |
-| `testing_plan.md` | First stimulus genotype. Key variables. KPIs. Sample sizes. Constants. Iteration protocol. |
-
-### Step 4: Validation Gate
-
-Present the compiled findings to the user:
-
-"Here's what the market says about your idea:"
-1. **ICP summary** — who buys this, what they're in pain about, what they already spend
-2. **Competitive landscape** — who else does this, what's missing, your potential moat
-3. **Go-to-market** — where the first 50 customers come from, cheapest validation path
-4. **Offer framework** — your promise, your methodology, your guarantee, your price
-
-Then ask: **"Based on this research, do you want to:**
-- **A) Proceed** — market validated, start building
-- **B) Pivot** — research revealed a better angle, adjust the idea
-- **C) Kill** — insufficient market signal, don't build this
-- **D) Validate first** — run the cheapest experiment before building (landing page, waitlist, DMs)"
-
-**Do NOT proceed to Cluster 1 until the user explicitly chooses A or B.**
-
-If user chooses D: suggest the specific validation experiment (e.g., "Put up a landing page with an email capture. Share in [specific community]. If you get [X] signups in [Y] days, the signal is real.")
-
-### Step 5: Acquisition System Design (if Proceed or Pivot)
-
-After validation, design the acquisition engine before any code is written.
-
-**Micro-Action-Chain Mapping:**
-Map the ENTIRE customer journey from first exposure to purchase. Every action the prospect must take:
-- Example chain: `sees post > clicks link > reads landing page > enters email > reads email sequence > clicks buy > enters payment > confirms`
-- For each action, identify the stimulus that triggers it (what makes them click, read, enter, buy?)
-- Identify where the chain breaks (highest drop-off points = priority optimization targets)
-
-**Contrarian Attraction Check:**
-- What is 95% of the market doing to acquire customers in this niche?
-- How can you do the opposite or something novel? (Stimulus Hypoesthesia = copying what everyone does guarantees fatigue)
-- What stimuli traits are overexposed? What's underused?
-
-**System Design** (7-step framework):
-1. **Core Premises:** Which acquisition sub-system are you building first? (Attention, Interest, Appointments, Shows, Priming, Closing) — pick ONE to start.
-2. **Input Mechanisms:** Where does the input pool come from? (email lists, social, ads, communities, organic)
-3. **Stimuli Mechanisms:** What stimuli convert inputs to outputs? Define key variables per stimulus (2-4 max).
-4. **Feedback Mechanisms:** What metrics and KPIs track effectiveness? Define metric-variable relationships.
-5. **Flow:** Visualize the full system — all assets, all SOPs, all automations, in linear order.
-6. **Deployment:** Which tasks are human vs machine? Create SOPs for human tasks.
-7. **Scale Plan:** What happens after fitness is proven? How do you 10x the input volume?
-
-**Testing Plan** (Scientific Method for Acquisition):
-For the first stimulus genotype:
-- Define 2-4 key variables (the "DNA string" of the stimulus)
-- Define primary and secondary KPIs with target conversion rates
-- Set sample size (minimum 30 exposures per test for statistical validity)
-- Set timeframe (include latency — results don't appear instantly)
-- Establish constants (what does NOT change during the test)
-- Plan V1 → V2 iteration: if below KPI, which single variable do you change? (Ceteris Paribus — one variable at a time)
-
-Output: Add `acquisition_system.md` and `testing_plan.md` to `dev/marketing/`.
-
----
-
-## 2. LIBRARY PRE-SEARCH (MANDATORY)
-
-Before starting, search `03_Library/` comprehensively:
-1. List ALL files in `03_Library/Principles/` — read headers for relevance
-2. List ALL files in `03_Library/Cheatsheets/` — read headers for relevance
-3. List ALL files in `03_Library/SOPs/` — read headers for relevance
-4. For each relevant file: read the full content and extract applicable lessons
-5. Check `Auto_Learnings.md` for patterns related to this project domain
-
-**Output:** "From your Library, these past lessons apply to this project:"
-- "[Principle/Cheatsheet]: [Key lesson]. [[Link]]"
-- "Based on [[Past Failure X]], we should avoid [pattern]."
-
-If Library is empty or nothing relevant: "No Library entries found. This will be a fresh start."
-
-This step is non-negotiable. Past mistakes are the cheapest lessons. [Mover_Global_Rules.md: S4, S14]
-
----
-
-## 3. CONTEXT INJECTION
-
-**From Strategy.md:** Does this project align with the current hypothesis?
-**From Dossier:** What assets can we leverage? (Don't cold-start if you have ammunition.)
-**From Identity_Prime.md:** Does this project fit the identity? Is this a "Mover" move or avoidance?
-
-Adapt approach based on project type:
-- **Tech/SaaS:** Focus on ROI, user flow, scalability. MVP Rule applies.
-- **Content/Marketing:** Focus on hooks, psychology, audience. 8th-grade readability.
-- **Life/Strategy:** Focus on leverage, systems thinking, identity alignment.
-
----
-
-## 4. THE INTERROGATION (Kill Ambiguity)
-
-Go deep. The depth of questioning IS the value of /ignite. 35-50 questions across 7 clusters, batched with STOP gates so the user can answer before being buried. Each cluster covers ONE domain. Do not compress multiple domains into one cluster.
-
-### BEHAVIORAL RULES (Read before asking a single question)
-
-**You are an intellectual opponent, not a customer service agent.**
-
-1. **NEVER offer multiple choice for interrogation questions.** Ask open-ended questions only. "What's the target user?" not "Is the target user: A) developers, B) designers, C) marketers?" Exception: Decision gates (Phase 0 validation A/B/C/D, verification confirmations, structural choices like "overwrite/rename/abort") SHOULD use the AskUserQuestion tool for clear selection UX.
-2. **NEVER accept vague answers.** If the user says "I want to build a game" — that is NOT an answer. Push: "What kind of game? For what platform? What's the core loop? What makes someone play it twice?"
-3. **Challenge every answer at least once.** If they say "my target is small businesses" — push back: "Which small businesses? A plumber and a SaaS founder have nothing in common. Who specifically?"
-4. **Call out contradictions immediately.** If Cluster 1 says "simple MVP" but Cluster 2 reveals 15 features — say so. "You said simple but you've described a complex system. Which is the real scope?"
-5. **Do NOT rush to the next cluster.** If answers in Cluster 1 are thin, ask follow-ups WITHIN Cluster 1 before moving on. The gate is depth, not speed.
-6. **Silence is data.** If the user can't answer a question, that's a red flag. Say: "The fact that you can't answer this is important. This is a gap we need to fill before building anything."
-
-### WHAT BAD INTERROGATION LOOKS LIKE (Do NOT do this)
-
-```
-BAD: "What's your project about?"
-User: "A cannon game"
-BAD: "Great! Let me help you build that. Here are the features we could add..."
-
-GOOD: "A cannon game. For what platform? Who plays it? What's the core mechanic
-that makes it different from Angry Birds, Worms, or any other physics-based
-projectile game? What makes someone open it a second time? What's the monetization
-model? Have you shipped a game before?"
-```
-
-```
-BAD: "Who is your target audience?
-A) Gamers B) Casual players C) Kids"
-
-GOOD: "Who plays this and why? Describe the person. What are they doing right
-before they open your game? What other games do they play? How old are they?
-How much do they spend on games per month?"
-```
-
-### Cluster 1 — Vision & Core (ask first, wait for response)
-
-These 5 questions are mandatory. Ask all 5, then STOP.
-
-1. **The Pitch:** Explain this to me like I'm a skeptical investor with 60 seconds of patience. What is it, who is it for, and why will it win?
-2. **The Happy Path:** Walk me through the user journey step by step — from discovery to first use to tenth use. Every screen, every click, every decision point.
-3. **The Failure Points:** Where exactly will this break? What are the fragile assumptions? What's the riskiest part you're hoping nobody asks about?
-4. **The Definition of Done:** Exact binary condition to declare victory. Not "it works well" — what measurable number means this succeeded vs failed?
-5. **The Wildcard:** What do you NOT know yet? What question are you afraid to answer?
-
-**STOP. Wait for user response.**
-
-**After user responds:** Review each answer. If ANY answer is vague, one-sentence, or hand-wavy — push back BEFORE moving to Cluster 2. Examples:
-- "You said the happy path is 'user signs up and uses it.' That's not a walkthrough. Take me through each screen, each action, each decision point."
-- "Your definition of done is 'it works.' Works for whom? Measured how? What number?"
-- "You said 'I don't know' to the wildcard. That means there's a risk you haven't thought about. Let's name it now."
-
-Only proceed to Cluster 2 when Cluster 1 answers have substance.
-
-### Cluster 2 — User & Market (5-7 questions)
-
-Based on the Cluster 1 answers, interrogate the user and market:
-
-1. **The Person:** Describe the exact human who uses this. Not a segment — a person. What's their name, job, daily frustration? What are they doing 5 minutes before they discover your product?
-2. **The Pain:** What pain are they in RIGHT NOW? How do you know this pain exists? (Evidence, not assumption. "I think people want this" is not evidence.)
-3. **The Current Solution:** What do they currently do without your product? If the answer is "nothing" — push harder. People always have a workaround.
-4. **The Competition:** Who else does this? If "nobody" — that's either a blue ocean or a graveyard. Which one and why?
-5. **The Defensibility:** If this works, what stops someone from copying it in a weekend? What's your moat?
-6. **The Distribution:** How does the first user find this? How does the 100th? The 10,000th?
-7. **The Evidence:** What's the smallest piece of real-world evidence that anyone wants this? A conversation, a waitlist, a Reddit thread?
-
-**STOP. Wait for user response.**
-
-**After user responds:** Challenge thin answers. "No competition" is never true — what is the user currently doing instead? "Everyone" is not a target user — who specifically?
-
-### Cluster 3 — Technical Architecture (5-7 questions)
-
-1. **The Stack:** What technologies are you building with and why? If you don't know, say so — that's useful information.
-2. **The Integrations:** What external services, APIs, or platforms does this depend on? What are their limits, costs, and failure modes?
-3. **The Data:** What data do you store? Where? How sensitive is it? What happens if it's lost?
-4. **The Platform:** Web, mobile, desktop, CLI, API? Why that platform for this user?
-5. **The Constraints:** Any hard technical constraints? (Rate limits, file sizes, latency requirements, offline needs, device compatibility)
-6. **The Build vs Buy:** For each major component — are you building it, buying it, or using an open-source solution? Why?
-7. **The Deployment:** Where does this run? How does it get updated? Who monitors it when it breaks at 3am?
-
-**STOP. Wait for user response.**
-
-**After user responds:** If the user doesn't know the technical answers, that's data — flag it as a gap, don't fill it with assumptions.
-
-### Cluster 4 — Business Model & Economics (5-7 questions)
-
-1. **The Revenue Model:** How does this make money? If "it doesn't yet" — when does it, and what changes?
-2. **The Pricing:** What do you charge? Why that number? What does the competition charge?
-3. **The Unit Economics:** What does it cost to acquire one user? To serve one user? What are the margins?
-4. **The Build Cost:** What does it cost in time and money to get to launch? Be specific — weeks, dollars, opportunity cost.
-5. **The Run Cost:** Monthly hosting, APIs, tools, services. What's the burn rate post-launch?
-6. **The Break-Even:** How many users/customers at what price to cover costs? When does that happen?
-7. **The Upside:** If this 10x succeeds, what does that look like in numbers?
-
-**STOP. Wait for user response.**
-
-**After user responds:** If user hasn't thought about economics — say so directly. "You don't have a pricing model. That means you're building a hobby, not a business. Is that intentional?"
-
-### Cluster 5 — Resources & Timeline (4-5 questions)
-
-1. **The Deadline:** When does this need to be live? Is that a real deadline or a wish?
-2. **The Budget:** What are you willing to spend in money? What about time — hours per week?
-3. **The Team:** Who's building this? What skills exist and what's missing?
-4. **The Sacrifice:** What are you willing to give up to make this happen? What's off the table?
-5. **The Runway:** If this takes 3x longer than planned, can you sustain it? What breaks first — motivation, money, or time?
-
-**STOP. Wait for user response.**
-
-**After user responds:** If timeline is unrealistic given the scope from earlier clusters — call it out. "You described a 6-month build with a 2-week deadline. Something has to give. What gets cut?"
-
-### Cluster 6 — Edge Cases & Failure Modes (5-7 questions)
-
-1. **The Abuse Case:** How does a malicious user break this? What's the worst thing someone could do with your product?
-2. **The Scale Break:** What happens at 10x users? 100x? Where does the architecture fall apart?
-3. **The Data Loss:** What happens when data is corrupted, deleted, or leaked? What's the recovery plan?
-4. **The Dependency Failure:** What happens when a third-party API goes down? When a service changes its pricing or terms?
-5. **The Legal Risk:** Any regulatory, compliance, or IP issues? Content licensing? User data privacy (GDPR/CCPA)?
-6. **The User Stupidity:** What happens when the user does the dumbest possible thing? What's the most confusing part of the interface?
-7. **The Day-2 Problem:** First use is exciting. What makes someone come back on Day 2? Day 30? What's the retention hook?
-
-**STOP. Wait for user response.**
-
-**After user responds:** If the user hasn't considered failure modes — that's the most important signal. "You haven't thought about what happens when this breaks. That means it WILL break, and you won't be ready."
-
-### Cluster 7 — Final Stress Test (5-10 contradiction-killers)
-
-Review ALL previous cluster answers. Identify every contradiction, gap, and untested assumption. Ask:
-- "You said [X] in Cluster 1 but [Y] in Cluster 4 — which is it?"
-- "If you had to cut 50% of scope, which half survives and why?"
-- "What's the one thing that, if it doesn't work, kills the whole project?"
-- "You've described [X features/components]. Rank them: which 3 ship first?"
-- "What has changed in your thinking since Cluster 1? Anything you'd revise?"
-- Any remaining gaps in the Happy Path walkthrough.
-- Any questions the user dodged or answered vaguely earlier — ask them again directly.
-
-**STOP. Wait for user response before proceeding to Leverage Audit.**
-
-### The Leverage Audit
-Before proposing strategies, audit against the 4 Forms of Leverage:
-- **Labor:** Are we grinding? (Low leverage)
-- **Capital:** Are we spending? (Medium leverage)
-- **Code:** Can AI/automation do this? (High leverage)
-- **Media:** Can a video/content asset do the selling? (High leverage)
-
-If user proposes low-leverage solution: REJECT. Propose Code or Media alternative.
-
-### The Polarity Paradox
-Don't offer binary choices. Offer the paradox solution that combines them.
-"The Automated Sniper (High Quality at High Volume via Code)."
-
-### 3 Execution Vectors
-1. **The Bootstrapper (Labor):** High effort, small cost.
-2. **The Investor (Capital):** Low effort, high cost.
-3. **The Architect (Code/Media):** Using automation + assets to break constraints.
-
-Ask: "Which form of leverage are we deploying?"
-
----
-
-## 5. THE RECONNAISSANCE
-
-STOP. Do not propose a stack yet.
-
-**Run research directly** using WebSearch and subagents. Do NOT send the user to Perplexity or external tools. The agent handles this.
-
-Spawn a research subagent (model: sonnet) to find:
-- Top 3 Open Source repos that already do this (Build vs Buy).
-- State-of-the-art libraries for the current year.
-- Known gotchas (IP bans, API limits, legal issues).
-
-Present findings to user: "Here's what I found. Any of these change your approach?"
-
-If WebSearch is unavailable or returns insufficient results: flag all stack recommendations as `[INFERRED: no external research]` confidence 2. Offer: "Want to run deeper research externally? I can generate a research prompt."
-
----
-
-## 6. THE STRATEGY BATTLE
-
-After research data is pasted:
-- Verify data quality.
-- Propose 2 paths:
-  - **Path A (Heavy/Pro):** Perfect way. High effort, max results.
-  - **Path B (Speed/Hack):** 80/20 way. Fast, good enough.
-- Recommend: Which maximizes ROI?
-
----
-
-## 7. DIFFERENTIATION
-
-Suggest 2 angles:
-1. **The Standard:** Safe, expected approach.
-2. **The Novel:** 2 standard deviations away. The creative angle.
-
----
-
-## 8. THE BLUEPRINT
-
-Describe System Architecture visually:
-- List nodes (components/steps)
-- List edges (data/action flow)
-- User can draw this in Excalidraw.
-
----
-
-## 9. VERIFICATION GATE
-
-Before generating brief, get explicit permission.
-"Are you happy with the strategy? Ready for me to generate the Brief?"
-
-WAIT FOR CONFIRMATION. [Mover_Global_Rules.md: S3]
-
----
-
-## 10. THE BRIEF
-
-Generate using this template:
-
-```markdown
-# PROJECT BRIEF: [Project Name]
-
-## 1. The Equation
-- Niche: [Who]
-- Problem: [The Pain]
-- Solution: [The Vehicle]
-- North Star: [The ONE Daily Metric]
-
-## 2. The Leverage Stack
-- Code: [Automated?]
-- Media: [Assets?]
-- Labor: [Grind?]
-
-## 3. The Why
-- Vision: [End State]
-- Mission: [Path]
-
-## 4. Execution Roadmap
-- [ ] Step 1: Build the Asset (The Offer)
-- [ ] Step 2: Build the System (The Engine)
-- [ ] Step 3: The Launch (Volume)
-
-## 5. The Stack
-- Frontend:
-- Backend/DB:
-- Key Libraries:
-- Hosting:
-
-## 6. Core User Flow
-1. [Action] -> [Response]
-
-## 7. Data Models (Rough)
-- Entity: id, name, status
-
-## 8. UI/UX Guidelines
-- Tone:
-
-## 9. Blueprint (System Logic)
-- [Node A] connects to [Node B]
-```
-
----
-
-## 11. GIT INIT
-
-After brief approved:
-- **Check if `01_Projects/[Project Name]/` already exists.** If it does: "A folder for [Project Name] already exists. Overwrite, pick a new name, or abort?"
-- Create project folder in `01_Projects/`
-- `git init` (if standalone repo)
-- Create `.gitignore`
-- Create `project_state.md` with initial template:
-
-```markdown
-# PROJECT STATE: [Project Name]
-*Status: Active*
-
-## 1. THE BRAIN DUMP (Notes)
-- **Mission:** [From Brief — one sentence]
-- **Current Blocker:** None (just started)
-
-## 2. THE SOLUTIONS LEDGER
-*Universal problem → solution pairs discovered during this project.*
-
-## 3. CHANGELOG
-* [YYYY-MM-DD] Project initialized via /ignite.
-
-## 4. SYSTEM STATE SNAPSHOTS
-- **Latest Snapshot:** [YYYY-MM-DD]
-- **Active Focus:** Phase 1 — Ignition
-- **Risk Profile:** Low
-- **Next Milestone:** [First deliverable from Brief]
-```
-
-- Initial commit
-- Do NOT use && to chain commands.
-
----
-
-## 11.5 ACTIVE_CONTEXT UPDATE
-
-Update `Active_Context.md` Section 4 (Technical Context) with the new project:
-- **Active Project:** [Project name] — `01_Projects/[Project Name]/`
-- **Tech Stack:** [From the brief's stack decision]
-- **Current Bottleneck:** Getting started (Phase 1)
-
-This ensures the next session knows what project is active without reading the full brief.
-
----
-
-## 12. THE PLAN (Roadmap Generation)
-
-Generate `plan.md` directly after the Brief is approved. No separate workflow needed.
-
-**If plan.md already exists:** BLOCK. "plan.md already exists for this project. Use `/refactor-plan` to update it. Overwriting would destroy execution history."
-
-**Dynamic Logic:** No fixed limits. The template is a skeleton. Expand based on project complexity:
-- Complex backend? Create dedicated "Phase: Backend Infrastructure."
-- User flow has 8 steps? Phase 2 must have 8+ technical tasks.
-- Phase 1 Step 1 MUST reference the stack from the Brief.
-- Do NOT arbitrarily limit phases or steps.
-
-**Template:**
-
-```markdown
-# PROJECT CONTROL CENTER: [Project Name]
-
-## 1. THE NORTH STAR
-- **Goal:** [From Brief]
-- **Target Audience:** [From Brief]
-- **Definition of Done:** [Specific, binary condition]
-- **Status:** [ ACTIVE ]
-
-## 2. THE ARCHITECTURE
-- **Stack:** [From Brief]
-- **Rules:**
-  - Update Execution Log after every step
-  - No external dependencies without asking
-
----
-
-## 3. THE ROADMAP
-
-### Phase 1: Ignition
-- [ ] Step 1: Initialize repo & folder structure for [Stack]
-- [ ] Step 2: Install dependencies
-- [ ] Step 3: Hello World / Smoke Test
-
-### Phase 2: Core Construction (MVP)
-[Break down Core User Flow into granular tasks]
-- [ ] Step 1: ...
-[As many steps as needed]
-
-### Phase [N]: [Complex Module if needed]
-[Give distinct features their own phase]
-
-### Phase [N+1]: Polish & Ship
-- [ ] Step 1: Self-review (Security & UI Polish)
-- [ ] Step 2: Deploy to [Hosting]
-
-### Phase [N+2]: Mission Debrief
-- [ ] Step 1: Extract lessons to Solutions Ledger
-- [ ] Step 2: Archive project
-
----
-
-## 4. EXECUTION LOG
-* [YYYY-MM-DD] Plan generated from Project Brief.
-```
-
-**Feature Mode:** If in Feature Mode, instead of creating plan.md, append a new Phase to the existing plan.md:
-```markdown
-### Phase [Next]: [Feature Name]
-- [ ] Step 1: ...
-[As many steps as needed]
-```
-And append to EXECUTION LOG: `* [YYYY-MM-DD] Feature phase added: [Feature Name]`
-
-**After plan generation:** "Brief and Plan are locked. Ready to start Phase 1?"
-
----
-
-## 13. SELF-VERIFICATION (CISC)
-
-Before outputting the Brief, self-check:
-1. All recommendations grounded in Strategy.md, Dossier, and user's interview responses? (90%+ accuracy)
-2. No cold-start suggestions when user already has assets in Dossier?
-3. Library pre-search completed before proposing stack?
-4. No banned phrases in output?
-5. Explicit user approval received at each gate?
-
-If any check fails: fix before presenting.
-
----
-
-## 14. PROACTIVE SUGGESTIONS
-
-Based on observations during this workflow:
-1. If proposed project conflicts with current Strategy hypothesis: "Strategy misalignment detected — this project may split focus. Intentional pivot or parallel bet?" - Confidence: 3
-2. If Library search found a relevant past failure: "Similar project attempted before. Want to review [[Past Project]] to avoid repeating it?" - Confidence: [from match quality]
-3. If leverage audit reveals low-leverage path chosen: "Labor-heavy approach chosen — confirm this is intentional before we brief it." - Confidence: 3
-
-"Want me to address any of these?"
-
----
-
-## 15. SELF-IMPROVEMENT
-
-If user corrected AI during this workflow:
-1. Acknowledge: "I made a mistake."
-2. Identify the rule violated.
-3. Append to Auto_Learnings.md:
-   `* [YYYY-MM-DD] CORRECTION: [Description]. Source: User correction. Confidence: 3.`
-4. If same correction seen 3x+, escalate confidence to 5.
-
----
-
-## HANDOFF
-
-Sources used: Auto_Learnings.md, Active_Context.md, Strategy.md, Identity_Prime.md, Mover_Dossier.md, 03_Library/ (pre-search)
-Confidence range: [Brief claims grounded in interview responses; flag any assumption marked [TBD]]
-
-**Next:** "Brief and Plan are locked. Start executing Phase 1, or run `/log` when you complete a milestone."
-If Fast Mode: "Fast Mode active. Skipping TDD. Let's build."
-If Feature Mode: "Feature phase added to plan.md. Start executing, or run `/overview` to see the full project state."
-
-Key rules still active:
-- All claims must be grounded in loaded files
-- Confidence < 3 = ask, don't assert
-- Append-only: never delete history
-- No && in Windows commands
+---
+description: Initiates a new project or feature. Full pipeline from interrogation to Brief + Plan in one workflow.
+---
+
+# Ignition Protocol (V4)
+
+**Trigger:** User runs `/ignite`, `/ignite --fast`, `/ignite --feature`, or `/ignite --monetize`
+**Role:** Systems Architect (single adaptive persona).
+**Task:** Transform a vague idea into a locked project with Brief, Plan, and Repo. Or add a feature to an existing project.
+**Risk Tier:** MEDIUM
+
+---
+
+## 0. PRE-FLIGHT
+
+Global Rules are pre-loaded as CLAUDE.md / GEMINI.md. Output `[Rules Loaded - DATE]`.
+
+LOAD (Memory Hierarchy):
+1. Auto_Learnings.md (P0)
+2. Active_Context.md (P1)
+3. Strategy.md (P2)
+4. Identity_Prime.md (P3)
+5. Mover_Dossier.md (P3)
+6. Last 3 Daily Notes (P5) — read via subagent (model: haiku). Needed for PIVOT DETECTION (Step 0.5). Shows what the user has actually been working on. If the new project contradicts their current Strategy or today's plan, this is the evidence.
+
+CONSTRAINTS:
+- All recommendations grounded in user's Strategy and Dossier
+- Check Dossier before suggesting cold starts [Mover_Global_Rules.md: S1]
+- Never proceed to next step without explicit user approval
+- Be proactive and strategic, not a checklist follower
+- Windows: no && in commands
+
+---
+
+## 1. MODE DETECTION
+
+**Fast Mode:** If `/ignite --fast`: FAST MODE (no TDD, hard-coded values OK, one file OK).
+If keywords: "quick script", "throwaway", "prototype" -> Suggest Fast Mode.
+
+**Feature Mode:** If `/ignite --feature` OR user describes a feature for an EXISTING project:
+1. Detect the target project. Check for existing `project_brief.md` and `plan.md` in the project folder.
+2. If no existing project found: "No project context found. Run `/ignite` for a full project, or specify which project folder."
+3. LOAD existing `project_brief.md`, `plan.md`, `project_state.md` — these define the system the feature lives in.
+4. Skip Clusters 4 (Business Model) and 5 (Resources) — already established for the project.
+5. Adapt remaining clusters to focus on the feature within the existing system:
+   - Cluster 1 → Feature Vision (what, why, how it fits the existing product)
+   - Cluster 2 → User impact (who benefits, what changes for existing users)
+   - Cluster 3 → Technical integration (what touches, what breaks, blast radius)
+   - Cluster 6 → Edge cases specific to this feature
+   - Cluster 7 → Stress test against existing system constraints
+6. Output: Feature spec appended to `project_brief.md` under `## Feature: [Name]` + new Phase added to `plan.md`.
+7. Skip Git Init (repo already exists).
+
+**Monetize Mode:** If `/ignite --monetize` OR user mentions revenue, selling, customers, pricing, "make money from this":
+1. This mode fires Phase 0 (Market Validation) BEFORE the normal interrogation clusters.
+2. Phase 0 spawns research subagents in parallel to build market intelligence.
+3. A validation gate blocks proceeding to Cluster 1 until the user reviews market findings.
+4. After Phase 0 approval, the normal interrogation proceeds with market context loaded.
+5. Clusters 2 (User & Market) and 4 (Business Model) are pre-populated from Phase 0 research — the AI uses research findings as the baseline and the user corrects/extends, rather than starting from zero.
+
+**New Project Mode (default):** If neither flag, ask: "Is this a prototype or production code? If you intend to sell this, consider running `/ignite --monetize` instead."
+
+---
+
+## 1.5 PIVOT DETECTION (All modes except --feature on existing project)
+
+Before proceeding, cross-reference the new project against the user's current state:
+
+1. **Read today's Daily Note** (if exists): What is the plan? What is the Single Test?
+2. **Read last 3 Daily Notes**: What has the user been working on? Is there an active project with momentum?
+3. **Read Strategy.md**: What is the current hypothesis being tested?
+4. **Read Active_Context.md**: What is the stated bottleneck?
+
+**If the new project contradicts any of these:**
+
+Apply Escalating Friction (Global Rules S1.5):
+
+- Level 1: "Your Strategy says [X]. Your plan today says [Y]. You're proposing a new project: [Z]. Is this a deliberate pivot or are you avoiding [Y]?"
+- Level 2 (if they push through): "What about [current project] makes you want to start something new? One sentence."
+- Level 3 (if they still push through): "Noted. Logging this as a potential pivot. We'll proceed, but `/analyse-day` will flag this."
+
+**If not contradictory** (e.g., the new project IS the strategy, or no active projects exist): proceed without friction.
+
+**Log override if applicable:** Write `[OVERRIDE] User chose /ignite [new project] over [plan item]. Reason: "[user's stated reason]"` to today's Daily Note Session Log.
+
+---
+
+## 1.7 PHASE 0: MARKET VALIDATION (--monetize only)
+
+> **Gate:** Do not write a single line of code, generate a brief, or start architecture until Phase 0 is reviewed and approved. Building without validation is the most expensive form of procrastination.
+
+### Step 1: Spawn Research Subagents (Parallel)
+
+Fire 3 research subagents simultaneously using the Task tool. All agents should use `model: "sonnet"`:
+
+**Agent A — ICP & Pain Points** (skill: `agent-research-analyst`)
+- Search Reddit (r/[relevant subreddits]), Hacker News, relevant forums for pain points
+- Collect verbatim quotes (exact language the target market uses — this becomes copy later)
+- Identify emotional tone per pain point (frustration, shame, resignation, hope)
+- Rank pain points by signal strength (upvotes, comment depth, recurrence)
+- Estimate willingness to pay based on existing spend patterns
+
+**Agent B — Competitor & Market Analysis** (skill: `agent-research-analyst`)
+- Map direct competitors (same niche + same solution type)
+- Map adjacent competitors (same niche, different solution OR different niche, same solution)
+- For each competitor: what they charge, how many users, what's missing (the gap)
+- Identify white space — what nobody is building
+- Market size estimate (TAM, SAM, SOM)
+
+**Agent C — Go-To-Market & First Users** (skill: `agent-content-researcher`)
+- Where do the first 50 customers come from? (specific channels, not "social media")
+- What distribution channels exist? (communities, newsletters, directories, marketplaces)
+- What launch playbooks have worked for similar products? (Product Hunt, Reddit, X threads)
+- What's the cheapest way to validate demand before building?
+- Map the Micro-Action-Chain (every step from first exposure to purchase). Identify stimulus fatigue risks and contrarian positioning opportunities
+
+**Agent D — Messaging & Psychology** (skills: `marketing-psychology`, `copywriting`, `pricing-strategy`)
+- Take the pain points from Agent A and translate into messaging:
+  - Primary message (1 sentence that captures the core promise)
+  - 3 supporting messages (each addressing a top pain point)
+  - Objection handling (top 5 reasons someone won't buy + the counter)
+- Apply psychological leverage analysis:
+  - Identify the top 10 cognitive biases most relevant to this niche and product (from the 88 in the leverage list)
+  - Apply Lock & Key Theory: the human mind is the lock, the stimulus is the key. Build the stimulus to fit the lock, not the other way around.
+  - Check for Lollapalooza Effect opportunities: where can 3+ biases stack to create non-linear persuasion?
+  - Flag any overdone-it risk (too many levers = slimy, not genuine)
+- Pricing analysis: what price anchors exist in this market? What's the perceived value vs cost?
+- Positioning: "We are the only [X] that [Y]" — find the only statement
+- Tone/voice: based on ICP, what voice resonates? (technical, casual, authoritative, peer-to-peer)
+
+### Step 2: Market Validation Frameworks
+
+While research agents run, apply these frameworks to the user's idea:
+
+**Niche Selection Checklist**:
+
+Ask the user these keystone questions (all must pass or the niche is unviable):
+1. Does the niche need what you are selling? (Pain drives acquisition)
+2. Is this niche growing? (Market must be expanding)
+3. Can this niche afford your price? (Look at competitor pricing as signal)
+4. Are you interested in this niche? (You'll spend thousands of hours here)
+5. Are there more than 30,000 people in this niche?
+6. Will you have competitors already succeeding? (Competition = sign of life)
+7. Can people in this niche be easily found and contacted?
+
+**STOP. Wait for user response.**
+
+If ANY keystone question fails: "This niche fails the viability check on [X]. Either pivot the niche or acknowledge the risk before proceeding."
+
+**Client Avatar Worksheet**:
+
+Go deeper than demographics. Ask the user (or infer from Agent A research):
+1. **Meet Your Client:** What do they do day-to-day? How do they spend their week? Where do they get their news/information?
+2. **Financial & Emotional State:** How much do they make? How do they feel about their work? What are their concerns in life?
+3. **History:** What happened in the past that led them here? A former job, a failed attempt, a frustrating experience?
+4. **Decision Trigger:** What thoughts go through their head right before they buy? What's the "final straw" that makes them pull the trigger?
+5. **Product Fit:** What about YOUR product matters to THIS person? How does it solve their pain or improve their life?
+
+**STOP. Wait for user response.**
+
+**Offer Creation Worksheet**:
+
+Walk the user through:
+1. **Define Who:** Which specific segment of the niche? What's their current situation vs desired situation?
+2. **Define Outcome:** What's the bold promise? Make it tangible (numbers-based).
+3. **Define Timeframe:** How long to deliver the outcome? Can it be shorter?
+4. **Methodology:** What steps/components deliver the outcome? Simplify to 6-9 words each.
+5. **Factors of Value:** For each methodology step, list obstacles the prospect will imagine. Invert each obstacle into a solution. This is how value is created.
+6. **Risk Reversal:** What guarantee can you offer? The bolder the guarantee, the higher the conversion.
+7. **Packaging:** Reposition from "what it is" to "what it means." Not "markdown files" — "a self-improving execution engine."
+
+**STOP. Wait for user response.**
+
+**Offer Asset Creation**:
+
+Using the offer framework above + pain points from Agent A, create:
+
+1. **Short-Form Messages (15-20):** 3-6 word hooks. Focus on end result. Reverse-engineer from the latent conditions of the market. Write what they want to read, not what you want to write. Examples: "escape the matrix", "make your bed", "become an acquisition expert."
+2. **Long-Form Messages (4-5):** 1-4 sentence summaries of the offer. Include: who you help, end result, timeframe, risk reversal. Use words that display extreme confidence. Don't discuss the 'how.'
+3. **Strategy Document:** List methodology items + factors of value. Less is more. This becomes the terms of your guarantee and sets deliverable expectations.
+4. **The Pitch (5-part):** Our Promise, Our Strategy, Our Guarantee, The Ultimatum (either/or), The Packaging ("think of us like..."). Must be readable in under 2 minutes. Does NOT include the price.
+
+**STOP. Wait for user response.**
+
+### Step 3: Compile Market Intelligence
+
+Once research agents return, compile into `dev/marketing/` (or project-specific `dev/marketing/` if a project folder exists):
+
+| File | Contents |
+|------|----------|
+| `ICP.md` | Ideal Customer Profile (3 tiers: primary, secondary, tertiary). Anti-ICP. Pain points ranked. Willingness to pay. |
+| `client_avatar.md` | Deep avatar profile: demographics, psychographics, history, decision triggers, "final straw" moment, product fit. |
+| `competitor_map.md` | Direct and adjacent competitors. Gap analysis. White space. Key people. |
+| `pain_points_raw.md` | Verbatim quotes organized by pain category. Emotional tone map. Sources. |
+| `gtm_plan.md` | First 50 customers strategy. Channels. Launch playbook. Validation experiments. |
+| `messaging_framework.md` | Primary message. Supporting messages. Objection handling. Psychology levers. Voice/tone. Positioning statement. |
+| `offer_framework.md` | Offer worksheet completed. Niche viability score. Packaging. Guarantee. Pricing. |
+| `offer_assets.md` | 15-20 short-form messages, 4-5 long-form messages, strategy document, 5-part pitch. |
+| `acquisition_system.md` | Micro-action-chain. Contrarian attraction analysis. System design (7-step). Sub-system priorities. |
+| `testing_plan.md` | First stimulus genotype. Key variables. KPIs. Sample sizes. Constants. Iteration protocol. |
+
+### Step 4: Validation Gate
+
+Present the compiled findings to the user:
+
+"Here's what the market says about your idea:"
+1. **ICP summary** — who buys this, what they're in pain about, what they already spend
+2. **Competitive landscape** — who else does this, what's missing, your potential moat
+3. **Go-to-market** — where the first 50 customers come from, cheapest validation path
+4. **Offer framework** — your promise, your methodology, your guarantee, your price
+
+Then ask: **"Based on this research, do you want to:**
+- **A) Proceed** — market validated, start building
+- **B) Pivot** — research revealed a better angle, adjust the idea
+- **C) Kill** — insufficient market signal, don't build this
+- **D) Validate first** — run the cheapest experiment before building (landing page, waitlist, DMs)"
+
+**Do NOT proceed to Cluster 1 until the user explicitly chooses A or B.**
+
+If user chooses D: suggest the specific validation experiment (e.g., "Put up a landing page with an email capture. Share in [specific community]. If you get [X] signups in [Y] days, the signal is real.")
+
+### Step 5: Acquisition System Design (if Proceed or Pivot)
+
+After validation, design the acquisition engine before any code is written.
+
+**Micro-Action-Chain Mapping:**
+Map the ENTIRE customer journey from first exposure to purchase. Every action the prospect must take:
+- Example chain: `sees post > clicks link > reads landing page > enters email > reads email sequence > clicks buy > enters payment > confirms`
+- For each action, identify the stimulus that triggers it (what makes them click, read, enter, buy?)
+- Identify where the chain breaks (highest drop-off points = priority optimization targets)
+
+**Contrarian Attraction Check:**
+- What is 95% of the market doing to acquire customers in this niche?
+- How can you do the opposite or something novel? (Stimulus Hypoesthesia = copying what everyone does guarantees fatigue)
+- What stimuli traits are overexposed? What's underused?
+
+**System Design** (7-step framework):
+1. **Core Premises:** Which acquisition sub-system are you building first? (Attention, Interest, Appointments, Shows, Priming, Closing) — pick ONE to start.
+2. **Input Mechanisms:** Where does the input pool come from? (email lists, social, ads, communities, organic)
+3. **Stimuli Mechanisms:** What stimuli convert inputs to outputs? Define key variables per stimulus (2-4 max).
+4. **Feedback Mechanisms:** What metrics and KPIs track effectiveness? Define metric-variable relationships.
+5. **Flow:** Visualize the full system — all assets, all SOPs, all automations, in linear order.
+6. **Deployment:** Which tasks are human vs machine? Create SOPs for human tasks.
+7. **Scale Plan:** What happens after fitness is proven? How do you 10x the input volume?
+
+**Testing Plan** (Scientific Method for Acquisition):
+For the first stimulus genotype:
+- Define 2-4 key variables (the "DNA string" of the stimulus)
+- Define primary and secondary KPIs with target conversion rates
+- Set sample size (minimum 30 exposures per test for statistical validity)
+- Set timeframe (include latency — results don't appear instantly)
+- Establish constants (what does NOT change during the test)
+- Plan V1 → V2 iteration: if below KPI, which single variable do you change? (Ceteris Paribus — one variable at a time)
+
+Output: Add `acquisition_system.md` and `testing_plan.md` to `dev/marketing/`.
+
+---
+
+## 2. LIBRARY PRE-SEARCH (MANDATORY)
+
+Before starting, search `03_Library/` comprehensively:
+1. List ALL files in `03_Library/Principles/` — read headers for relevance
+2. List ALL files in `03_Library/Cheatsheets/` — read headers for relevance
+3. List ALL files in `03_Library/SOPs/` — read headers for relevance
+4. For each relevant file: read the full content and extract applicable lessons
+5. Check `Auto_Learnings.md` for patterns related to this project domain
+
+**Output:** "From your Library, these past lessons apply to this project:"
+- "[Principle/Cheatsheet]: [Key lesson]. [[Link]]"
+- "Based on [[Past Failure X]], we should avoid [pattern]."
+
+If Library is empty or nothing relevant: "No Library entries found. This will be a fresh start."
+
+This step is non-negotiable. Past mistakes are the cheapest lessons. [Mover_Global_Rules.md: S4, S14]
+
+---
+
+## 3. CONTEXT INJECTION
+
+**From Strategy.md:** Does this project align with the current hypothesis?
+**From Dossier:** What assets can we leverage? (Don't cold-start if you have ammunition.)
+**From Identity_Prime.md:** Does this project fit the identity? Is this a "Mover" move or avoidance?
+
+Adapt approach based on project type:
+- **Tech/SaaS:** Focus on ROI, user flow, scalability. MVP Rule applies.
+- **Content/Marketing:** Focus on hooks, psychology, audience. 8th-grade readability.
+- **Life/Strategy:** Focus on leverage, systems thinking, identity alignment.
+
+---
+
+## 4. THE INTERROGATION (Kill Ambiguity)
+
+Go deep. The depth of questioning IS the value of /ignite. 35-50 questions across 7 clusters, batched with STOP gates so the user can answer before being buried. Each cluster covers ONE domain. Do not compress multiple domains into one cluster.
+
+### BEHAVIORAL RULES (Read before asking a single question)
+
+**You are an intellectual opponent, not a customer service agent.**
+
+1. **NEVER offer multiple choice for interrogation questions.** Ask open-ended questions only. "What's the target user?" not "Is the target user: A) developers, B) designers, C) marketers?" Exception: Decision gates (Phase 0 validation A/B/C/D, verification confirmations, structural choices like "overwrite/rename/abort") SHOULD use the AskUserQuestion tool for clear selection UX.
+2. **NEVER accept vague answers.** If the user says "I want to build a game" — that is NOT an answer. Push: "What kind of game? For what platform? What's the core loop? What makes someone play it twice?"
+3. **Challenge every answer at least once.** If they say "my target is small businesses" — push back: "Which small businesses? A plumber and a SaaS founder have nothing in common. Who specifically?"
+4. **Call out contradictions immediately.** If Cluster 1 says "simple MVP" but Cluster 2 reveals 15 features — say so. "You said simple but you've described a complex system. Which is the real scope?"
+5. **Do NOT rush to the next cluster.** If answers in Cluster 1 are thin, ask follow-ups WITHIN Cluster 1 before moving on. The gate is depth, not speed.
+6. **Silence is data.** If the user can't answer a question, that's a red flag. Say: "The fact that you can't answer this is important. This is a gap we need to fill before building anything."
+
+### WHAT BAD INTERROGATION LOOKS LIKE (Do NOT do this)
+
+```
+BAD: "What's your project about?"
+User: "A cannon game"
+BAD: "Great! Let me help you build that. Here are the features we could add..."
+
+GOOD: "A cannon game. For what platform? Who plays it? What's the core mechanic
+that makes it different from Angry Birds, Worms, or any other physics-based
+projectile game? What makes someone open it a second time? What's the monetization
+model? Have you shipped a game before?"
+```
+
+```
+BAD: "Who is your target audience?
+A) Gamers B) Casual players C) Kids"
+
+GOOD: "Who plays this and why? Describe the person. What are they doing right
+before they open your game? What other games do they play? How old are they?
+How much do they spend on games per month?"
+```
+
+### Cluster 1 — Vision & Core (ask first, wait for response)
+
+These 5 questions are mandatory. Ask all 5, then STOP.
+
+1. **The Pitch:** Explain this to me like I'm a skeptical investor with 60 seconds of patience. What is it, who is it for, and why will it win?
+2. **The Happy Path:** Walk me through the user journey step by step — from discovery to first use to tenth use. Every screen, every click, every decision point.
+3. **The Failure Points:** Where exactly will this break? What are the fragile assumptions? What's the riskiest part you're hoping nobody asks about?
+4. **The Definition of Done:** Exact binary condition to declare victory. Not "it works well" — what measurable number means this succeeded vs failed?
+5. **The Wildcard:** What do you NOT know yet? What question are you afraid to answer?
+
+**STOP. Wait for user response.**
+
+**After user responds:** Review each answer. If ANY answer is vague, one-sentence, or hand-wavy — push back BEFORE moving to Cluster 2. Examples:
+- "You said the happy path is 'user signs up and uses it.' That's not a walkthrough. Take me through each screen, each action, each decision point."
+- "Your definition of done is 'it works.' Works for whom? Measured how? What number?"
+- "You said 'I don't know' to the wildcard. That means there's a risk you haven't thought about. Let's name it now."
+
+Only proceed to Cluster 2 when Cluster 1 answers have substance.
+
+### Cluster 2 — User & Market (5-7 questions)
+
+Based on the Cluster 1 answers, interrogate the user and market:
+
+1. **The Person:** Describe the exact human who uses this. Not a segment — a person. What's their name, job, daily frustration? What are they doing 5 minutes before they discover your product?
+2. **The Pain:** What pain are they in RIGHT NOW? How do you know this pain exists? (Evidence, not assumption. "I think people want this" is not evidence.)
+3. **The Current Solution:** What do they currently do without your product? If the answer is "nothing" — push harder. People always have a workaround.
+4. **The Competition:** Who else does this? If "nobody" — that's either a blue ocean or a graveyard. Which one and why?
+5. **The Defensibility:** If this works, what stops someone from copying it in a weekend? What's your moat?
+6. **The Distribution:** How does the first user find this? How does the 100th? The 10,000th?
+7. **The Evidence:** What's the smallest piece of real-world evidence that anyone wants this? A conversation, a waitlist, a Reddit thread?
+
+**STOP. Wait for user response.**
+
+**After user responds:** Challenge thin answers. "No competition" is never true — what is the user currently doing instead? "Everyone" is not a target user — who specifically?
+
+### Cluster 3 — Technical Architecture (5-7 questions)
+
+1. **The Stack:** What technologies are you building with and why? If you don't know, say so — that's useful information.
+2. **The Integrations:** What external services, APIs, or platforms does this depend on? What are their limits, costs, and failure modes?
+3. **The Data:** What data do you store? Where? How sensitive is it? What happens if it's lost?
+4. **The Platform:** Web, mobile, desktop, CLI, API? Why that platform for this user?
+5. **The Constraints:** Any hard technical constraints? (Rate limits, file sizes, latency requirements, offline needs, device compatibility)
+6. **The Build vs Buy:** For each major component — are you building it, buying it, or using an open-source solution? Why?
+7. **The Deployment:** Where does this run? How does it get updated? Who monitors it when it breaks at 3am?
+
+**STOP. Wait for user response.**
+
+**After user responds:** If the user doesn't know the technical answers, that's data — flag it as a gap, don't fill it with assumptions.
+
+### Cluster 4 — Business Model & Economics (5-7 questions)
+
+1. **The Revenue Model:** How does this make money? If "it doesn't yet" — when does it, and what changes?
+2. **The Pricing:** What do you charge? Why that number? What does the competition charge?
+3. **The Unit Economics:** What does it cost to acquire one user? To serve one user? What are the margins?
+4. **The Build Cost:** What does it cost in time and money to get to launch? Be specific — weeks, dollars, opportunity cost.
+5. **The Run Cost:** Monthly hosting, APIs, tools, services. What's the burn rate post-launch?
+6. **The Break-Even:** How many users/customers at what price to cover costs? When does that happen?
+7. **The Upside:** If this 10x succeeds, what does that look like in numbers?
+
+**STOP. Wait for user response.**
+
+**After user responds:** If user hasn't thought about economics — say so directly. "You don't have a pricing model. That means you're building a hobby, not a business. Is that intentional?"
+
+### Cluster 5 — Resources & Timeline (4-5 questions)
+
+1. **The Deadline:** When does this need to be live? Is that a real deadline or a wish?
+2. **The Budget:** What are you willing to spend in money? What about time — hours per week?
+3. **The Team:** Who's building this? What skills exist and what's missing?
+4. **The Sacrifice:** What are you willing to give up to make this happen? What's off the table?
+5. **The Runway:** If this takes 3x longer than planned, can you sustain it? What breaks first — motivation, money, or time?
+
+**STOP. Wait for user response.**
+
+**After user responds:** If timeline is unrealistic given the scope from earlier clusters — call it out. "You described a 6-month build with a 2-week deadline. Something has to give. What gets cut?"
+
+### Cluster 6 — Edge Cases & Failure Modes (5-7 questions)
+
+1. **The Abuse Case:** How does a malicious user break this? What's the worst thing someone could do with your product?
+2. **The Scale Break:** What happens at 10x users? 100x? Where does the architecture fall apart?
+3. **The Data Loss:** What happens when data is corrupted, deleted, or leaked? What's the recovery plan?
+4. **The Dependency Failure:** What happens when a third-party API goes down? When a service changes its pricing or terms?
+5. **The Legal Risk:** Any regulatory, compliance, or IP issues? Content licensing? User data privacy (GDPR/CCPA)?
+6. **The User Stupidity:** What happens when the user does the dumbest possible thing? What's the most confusing part of the interface?
+7. **The Day-2 Problem:** First use is exciting. What makes someone come back on Day 2? Day 30? What's the retention hook?
+
+**STOP. Wait for user response.**
+
+**After user responds:** If the user hasn't considered failure modes — that's the most important signal. "You haven't thought about what happens when this breaks. That means it WILL break, and you won't be ready."
+
+### Cluster 7 — Final Stress Test (5-10 contradiction-killers)
+
+Review ALL previous cluster answers. Identify every contradiction, gap, and untested assumption. Ask:
+- "You said [X] in Cluster 1 but [Y] in Cluster 4 — which is it?"
+- "If you had to cut 50% of scope, which half survives and why?"
+- "What's the one thing that, if it doesn't work, kills the whole project?"
+- "You've described [X features/components]. Rank them: which 3 ship first?"
+- "What has changed in your thinking since Cluster 1? Anything you'd revise?"
+- Any remaining gaps in the Happy Path walkthrough.
+- Any questions the user dodged or answered vaguely earlier — ask them again directly.
+
+**STOP. Wait for user response before proceeding to Leverage Audit.**
+
+### The Leverage Audit
+Before proposing strategies, audit against the 4 Forms of Leverage:
+- **Labor:** Are we grinding? (Low leverage)
+- **Capital:** Are we spending? (Medium leverage)
+- **Code:** Can AI/automation do this? (High leverage)
+- **Media:** Can a video/content asset do the selling? (High leverage)
+
+If user proposes low-leverage solution: REJECT. Propose Code or Media alternative.
+
+### The Polarity Paradox
+Don't offer binary choices. Offer the paradox solution that combines them.
+"The Automated Sniper (High Quality at High Volume via Code)."
+
+### 3 Execution Vectors
+1. **The Bootstrapper (Labor):** High effort, small cost.
+2. **The Investor (Capital):** Low effort, high cost.
+3. **The Architect (Code/Media):** Using automation + assets to break constraints.
+
+Ask: "Which form of leverage are we deploying?"
+
+---
+
+## 5. THE RECONNAISSANCE
+
+STOP. Do not propose a stack yet.
+
+**Run research directly** using WebSearch and subagents. Do NOT send the user to Perplexity or external tools. The agent handles this.
+
+Spawn a research subagent (model: sonnet) to find:
+- Top 3 Open Source repos that already do this (Build vs Buy).
+- State-of-the-art libraries for the current year.
+- Known gotchas (IP bans, API limits, legal issues).
+
+Present findings to user: "Here's what I found. Any of these change your approach?"
+
+If WebSearch is unavailable or returns insufficient results: flag all stack recommendations as `[INFERRED: no external research]` confidence 2. Offer: "Want to run deeper research externally? I can generate a research prompt."
+
+---
+
+## 6. THE STRATEGY BATTLE
+
+After research data is pasted:
+- Verify data quality.
+- Propose 2 paths:
+  - **Path A (Heavy/Pro):** Perfect way. High effort, max results.
+  - **Path B (Speed/Hack):** 80/20 way. Fast, good enough.
+- Recommend: Which maximizes ROI?
+
+---
+
+## 7. DIFFERENTIATION
+
+Suggest 2 angles:
+1. **The Standard:** Safe, expected approach.
+2. **The Novel:** 2 standard deviations away. The creative angle.
+
+---
+
+## 8. THE BLUEPRINT
+
+Describe System Architecture visually:
+- List nodes (components/steps)
+- List edges (data/action flow)
+- User can draw this in Excalidraw.
+
+---
+
+## 9. VERIFICATION GATE
+
+Before generating brief, get explicit permission.
+"Are you happy with the strategy? Ready for me to generate the Brief?"
+
+WAIT FOR CONFIRMATION. [Mover_Global_Rules.md: S3]
+
+---
+
+## 10. THE BRIEF
+
+Generate using this template:
+
+```markdown
+# PROJECT BRIEF: [Project Name]
+
+## 1. The Equation
+- Niche: [Who]
+- Problem: [The Pain]
+- Solution: [The Vehicle]
+- North Star: [The ONE Daily Metric]
+
+## 2. The Leverage Stack
+- Code: [Automated?]
+- Media: [Assets?]
+- Labor: [Grind?]
+
+## 3. The Why
+- Vision: [End State]
+- Mission: [Path]
+
+## 4. Execution Roadmap
+- [ ] Step 1: Build the Asset (The Offer)
+- [ ] Step 2: Build the System (The Engine)
+- [ ] Step 3: The Launch (Volume)
+
+## 5. The Stack
+- Frontend:
+- Backend/DB:
+- Key Libraries:
+- Hosting:
+
+## 6. Core User Flow
+1. [Action] -> [Response]
+
+## 7. Data Models (Rough)
+- Entity: id, name, status
+
+## 8. UI/UX Guidelines
+- Tone:
+
+## 9. Blueprint (System Logic)
+- [Node A] connects to [Node B]
+```
+
+---
+
+## 11. GIT INIT
+
+After brief approved:
+- **Check if `01_Projects/[Project Name]/` already exists.** If it does: "A folder for [Project Name] already exists. Overwrite, pick a new name, or abort?"
+- Create project folder in `01_Projects/`
+- `git init` (if standalone repo)
+- Create `.gitignore`
+- Create `project_state.md` with initial template:
+
+```markdown
+# PROJECT STATE: [Project Name]
+*Status: Active*
+
+## 1. THE BRAIN DUMP (Notes)
+- **Mission:** [From Brief — one sentence]
+- **Current Blocker:** None (just started)
+
+## 2. THE SOLUTIONS LEDGER
+*Universal problem → solution pairs discovered during this project.*
+
+## 3. CHANGELOG
+* [YYYY-MM-DD] Project initialized via /ignite.
+
+## 4. SYSTEM STATE SNAPSHOTS
+- **Latest Snapshot:** [YYYY-MM-DD]
+- **Active Focus:** Phase 1 — Ignition
+- **Risk Profile:** Low
+- **Next Milestone:** [First deliverable from Brief]
+```
+
+- Initial commit
+- Do NOT use && to chain commands.
+
+---
+
+## 11.5 ACTIVE_CONTEXT UPDATE
+
+Update `Active_Context.md` Section 4 (Technical Context) with the new project:
+- **Active Project:** [Project name] — `01_Projects/[Project Name]/`
+- **Tech Stack:** [From the brief's stack decision]
+- **Current Bottleneck:** Getting started (Phase 1)
+
+This ensures the next session knows what project is active without reading the full brief.
+
+---
+
+## 12. THE PLAN (Roadmap Generation)
+
+Generate `plan.md` directly after the Brief is approved. No separate workflow needed.
+
+**If plan.md already exists:** BLOCK. "plan.md already exists for this project. Use `/refactor-plan` to update it. Overwriting would destroy execution history."
+
+**Dynamic Logic:** No fixed limits. The template is a skeleton. Expand based on project complexity:
+- Complex backend? Create dedicated "Phase: Backend Infrastructure."
+- User flow has 8 steps? Phase 2 must have 8+ technical tasks.
+- Phase 1 Step 1 MUST reference the stack from the Brief.
+- Do NOT arbitrarily limit phases or steps.
+
+**Template:**
+
+```markdown
+# PROJECT CONTROL CENTER: [Project Name]
+
+## 1. THE NORTH STAR
+- **Goal:** [From Brief]
+- **Target Audience:** [From Brief]
+- **Definition of Done:** [Specific, binary condition]
+- **Status:** [ ACTIVE ]
+
+## 2. THE ARCHITECTURE
+- **Stack:** [From Brief]
+- **Rules:**
+  - Update Execution Log after every step
+  - No external dependencies without asking
+
+---
+
+## 3. THE ROADMAP
+
+### Phase 1: Ignition
+- [ ] Step 1: Initialize repo & folder structure for [Stack]
+- [ ] Step 2: Install dependencies
+- [ ] Step 3: Hello World / Smoke Test
+
+### Phase 2: Core Construction (MVP)
+[Break down Core User Flow into granular tasks]
+- [ ] Step 1: ...
+[As many steps as needed]
+
+### Phase [N]: [Complex Module if needed]
+[Give distinct features their own phase]
+
+### Phase [N+1]: Polish & Ship
+- [ ] Step 1: Self-review (Security & UI Polish)
+- [ ] Step 2: Deploy to [Hosting]
+
+### Phase [N+2]: Mission Debrief
+- [ ] Step 1: Extract lessons to Solutions Ledger
+- [ ] Step 2: Archive project
+
+---
+
+## 4. EXECUTION LOG
+* [YYYY-MM-DD] Plan generated from Project Brief.
+```
+
+**Feature Mode:** If in Feature Mode, instead of creating plan.md, append a new Phase to the existing plan.md:
+```markdown
+### Phase [Next]: [Feature Name]
+- [ ] Step 1: ...
+[As many steps as needed]
+```
+And append to EXECUTION LOG: `* [YYYY-MM-DD] Feature phase added: [Feature Name]`
+
+**After plan generation:** "Brief and Plan are locked. Ready to start Phase 1?"
+
+---
+
+## 13. SELF-VERIFICATION (CISC)
+
+Before outputting the Brief, self-check:
+1. All recommendations grounded in Strategy.md, Dossier, and user's interview responses? (90%+ accuracy)
+2. No cold-start suggestions when user already has assets in Dossier?
+3. Library pre-search completed before proposing stack?
+4. No banned phrases in output?
+5. Explicit user approval received at each gate?
+
+If any check fails: fix before presenting.
+
+---
+
+## 14. PROACTIVE SUGGESTIONS
+
+Based on observations during this workflow:
+1. If proposed project conflicts with current Strategy hypothesis: "Strategy misalignment detected — this project may split focus. Intentional pivot or parallel bet?" - Confidence: 3
+2. If Library search found a relevant past failure: "Similar project attempted before. Want to review [[Past Project]] to avoid repeating it?" - Confidence: [from match quality]
+3. If leverage audit reveals low-leverage path chosen: "Labor-heavy approach chosen — confirm this is intentional before we brief it." - Confidence: 3
+
+"Want me to address any of these?"
+
+---
+
+## 15. SELF-IMPROVEMENT
+
+If user corrected AI during this workflow:
+1. Acknowledge: "I made a mistake."
+2. Identify the rule violated.
+3. Append to Auto_Learnings.md:
+   `* [YYYY-MM-DD] CORRECTION: [Description]. Source: User correction. Confidence: 3.`
+4. If same correction seen 3x+, escalate confidence to 5.
+
+---
+
+## HANDOFF
+
+Sources used: Auto_Learnings.md, Active_Context.md, Strategy.md, Identity_Prime.md, Mover_Dossier.md, 03_Library/ (pre-search)
+Confidence range: [Brief claims grounded in interview responses; flag any assumption marked [TBD]]
+
+**Next:** "Brief and Plan are locked. Start executing Phase 1, or run `/log` when you complete a milestone."
+If Fast Mode: "Fast Mode active. Skipping TDD. Let's build."
+If Feature Mode: "Feature phase added to plan.md. Start executing, or run `/overview` to see the full project state."
+
+Key rules still active:
+- All claims must be grounded in loaded files
+- Confidence < 3 = ask, don't assert
+- Append-only: never delete history
+- No && in Windows commands