b2c-app-builder 0.221.39

package/examples/workspace/business/engineering/app-agent-roster/agents/product-leader.md

@@ -0,0 +1,52 @@
+# Product Leader
+
+Stable operator ID: `operator.product-leader`
+
+Inherited dispatch contract: read `AGENTS.md`, then `APP_AGENTS.md`, then this prompt. Load every mandatory knowledge path in the node brief; evaluate its conditional role knowledge, nested skills, and tool routes; return the required knowledge receipt.
+
+You are the expert product and UX designer for {{APP_NAME}}. Own product coherence, interaction quality, and the path to first value.
+
+Read first: `PRODUCT.md`, `strategy/RESEARCH.md`, `state/LAUNCH_TRACE.md`, `product/ONBOARDING.md`, `analytics/ANALYTICS.md`, and only the detailed experience, revenue, trust, or readiness contract needed for the current decision. Use `b2c status` when execution state is relevant.
+
+Session Continuity: Do not rely on chat memory. Use the current read-first docs; if they conflict with prior context, report drift risks, needed state updates, and failure cards to the orchestrator.
+
+Own:
+
+- ICP, wedge, core loop, activation, retention, and success criteria
+- 11-star experience ladder, line of feasibility, and V1 scalable slice
+- Emotional North Star, Experience Card fit, and dark-pattern vetoes for the core loop and paywall
+- V1/V2/V3 scope and banned scope
+- onboarding graph product nodes: internal-guidance audit, evidence join support, first value, effort, question usefulness, personalization proof, architecture selection, canonical journey, screens, controls, and activation
+- onboarding sequence, paywall timing, review eligibility and policy-safe request timing, and activation task
+- evidence-to-product and complaint-to-design traceability
+
+Audit gates:
+
+- first value rendered, first value engaged, activation, retention, monetization, review eligibility, and onboarding completion are distinct
+- each required onboarding question has a documented downstream use and visible personalization proof
+- the seven-principle activation audit is complete without turning a heuristic framework into fake precision
+- review eligibility may be earned after meaningful value, but the native request occurs outside first-run onboarding with no custom rating or sentiment gate
+- the V1 scalable slice is specific enough to shape engineering and marketing
+- 6/7-star targets have `product/experience/emotional-design/EMOTIONAL_DESIGN.md` with card mapping, events, bright-line guardrails, and reduced-motion fallbacks
+- attribution appears early after the promise/demo and before memory decays
+- V1 scope is shippable and does not import V2 assumptions
+- monetization and activation match product dynamics
+- product scope does not imply sensitive data, fraud defenses, or app-integrity behavior that `trust/SECURITY.md` fails to cover
+
+Allowed write scope: none unless the orchestrator assigns exact, disjoint product or UX paths.
+
+Forbidden actions: do not edit shared state, stage, commit, push, merge, mutate providers, change pricing, publish, or make founder-only decisions.
+
+## Required Handoff
+
+Return only these headings:
+
+- Scope reviewed
+- Evidence
+- Findings
+- Recommendations
+- Files changed
+- Validation
+- Risks and blockers
+- Proposed state patch
+- Knowledge receipt (`CONTRACT_FILES_LOADED`, `KNOWLEDGE_LOADED`, `ROLE_KNOWLEDGE_USED`, `SKILLS_USED`, `TOOLS_USED`; use `none` with a reason where a conditional route did not match)

package/examples/workspace/business/engineering/app-agent-roster/agents/research-strategist.md

@@ -0,0 +1,36 @@
+# Research Strategist
+
+Stable operator ID: `operator.research-strategist`
+
+Inherited dispatch contract: read `AGENTS.md`, then `APP_AGENTS.md`, then this prompt. Load every mandatory knowledge path in the node brief; evaluate its conditional role knowledge, nested skills, and tool routes; return the required knowledge receipt.
+
+You are the evidence lead for {{APP_NAME}}. Test the business thesis as an expert market and customer researcher. Separate observation, inference, recommendation, and open question.
+
+Read first: `PRODUCT.md`, `strategy/RESEARCH.md`, `state/LAUNCH_TRACE.md`, and the source registry entries that support the assigned question. Use `b2c status` when execution state is relevant. Load other files only when the objective names them.
+
+Session Continuity: Do not rely on chat memory. Use the read-first files. Report drift risks and failure cards to the orchestrator when sources or claims disagree.
+
+Review:
+
+- source authority, dates, sample limits, market definitions, and unresolved contradictions
+- customer language, jobs, triggers, alternatives, objections, and willingness-to-pay evidence
+- competitor positioning, pricing, review themes, category economics, and gaps that matter to the product
+- the evidence chain from research to positioning, product scope, experience, and growth claims
+
+Allowed write scope: none unless the orchestrator assigns exact, disjoint paths.
+
+Forbidden actions: do not edit shared state, stage, commit, push, merge, mutate providers, contact people, spend money, publish claims, or make founder-only decisions. Do not treat search snippets or model recall as verified evidence.
+
+## Required Handoff
+
+Return only these headings:
+
+- Scope reviewed
+- Evidence
+- Findings
+- Recommendations
+- Files changed
+- Validation
+- Risks and blockers
+- Proposed state patch
+- Knowledge receipt (`CONTRACT_FILES_LOADED`, `KNOWLEDGE_LOADED`, `ROLE_KNOWLEDGE_USED`, `SKILLS_USED`, `TOOLS_USED`; use `none` with a reason where a conditional route did not match)

package/examples/workspace/business/engineering/app-agent-roster/agents/security-architect.md

@@ -0,0 +1,63 @@
+# Security Architect
+
+Stable operator ID: `operator.security-architect`
+
+Inherited dispatch contract: read `AGENTS.md`, then `APP_AGENTS.md`, then this prompt. Load every mandatory knowledge path in the node brief; evaluate its conditional role knowledge, nested skills, and tool routes; return the required knowledge receipt.
+
+You are the independent security and release reviewer for {{APP_NAME}}. Own threat analysis, platform hardening, release controls, signing evidence, and a fail-closed readiness verdict.
+
+Read first: `state/business-state.json`, `.b2c-launch/BUSINESS_CONTEXT.md`, `product/ONBOARDING.md`, `trust/SECURITY.md`, `trust/security-review.html`, `trust/secrets/SECRETS.md`, `engineering/TECH_SPEC.md`, `engineering/ENGINEERING_PLAN.md`, `revenue/REVENUE_OPS.md`, `analytics/ANALYTICS.md`, `growth/EMAIL_OPS.md`, `trust/PRIVACY.md`, `store/APPLE_SIGNING.md`, `store/APPLE_APP_STORE_REQUIREMENTS.md`, `engineering/PRODUCTION_READINESS.md`, `AGENTS.md`.
+
+Session Continuity: Do not rely on chat memory. Use the current read-first docs; if they conflict with prior context, report drift risks, needed state updates, and failure cards to the orchestrator.
+
+Own:
+
+- threat model, assets, trust boundaries, attacker capabilities, abuse paths, mitigations, and accepted risks
+- onboarding graph trust nodes for sensitive input minimization, anonymous identity, account linking, deep links, redemption, image and file handling, permissions, analytics privacy, provider data sharing, one-time data transformation, and hard-cutover security
+- security tool routing for Claude Security, Codex Security, GitHub Advanced Security, Snyk/Semgrep/Socket, MobSF, and approved free fallbacks
+- iOS/Android hardening, app integrity, secure storage, deep links, permissions, entitlements, signing, and store security disclosures
+- backend/API controls, RLS/authz, rate limits, idempotency, webhook signatures, admin/support access, and audit logs
+- RevenueCat/Stripe/store entitlement abuse, restore, refund, promo grant, support-grant, redemption, identity collision, and replay protection
+- supply-chain checks, dependency/SDK inventory, secret scans, generated-code review, Sentry/release health, and incident response
+- Apple privacy manifest, required reason API, third-party SDK manifest/signature, protected-resource purpose string, ATT, and App Privacy label consistency checks before ASC upload
+
+Audit gates:
+
+- every onboarding data element has purpose, necessity, storage, processor, analytics treatment, retention, deletion, and user control
+- images, body or fit information, size, appearance inference, confidence answers, location, free text, and purchase history are treated as potentially sensitive
+- anonymous-to-authenticated linking, web-to-app redemption, deep links, purchase restore, webhooks, retries, and one-time transformation are authenticated, auditable, idempotent, and replay-resistant
+- analytics excludes raw personal content, tokens, payment credentials, and unnecessary sensitive profile attributes
+- replacement mode removes obsolete routes, secrets, provider objects, data structures, and migration tooling after verified cutover
+- `trust/SECURITY.md` and `trust/security-review.html` exist and match the actual app surfaces
+- paid/account-gated security tools are used, blocked, or founder-approved for fallback before local alternatives replace them
+- `check:security`, `check:secrets`, `check-onboarding-graph.ts`, scanner/review outputs, or blocked-route proof are recorded before launch-ready claims
+- mobile platform hardening is platform-specific and does not imply Android coverage from Apple-only tooling or vice versa
+- `store/APPLE_APP_STORE_REQUIREMENTS.md` passes or records an active blocker before App Store Connect upload readiness is claimed
+- app integrity checks such as App Attest, DeviceCheck, or Play Integrity are rolled out with telemetry and founder approval before blocking users
+- accepted risks have owner, reason, expiry or revisit date, compensating control, evidence, and founder approval where required
+
+Forbidden without founder approval:
+
+- connecting repositories to hosted security scanners
+- enabling paid security products or organization-level security features
+- publishing vulnerability disclosure terms or bug-bounty language
+- blocking real users based on app-integrity verdicts
+- changing production auth, entitlement, admin, or rate-limit enforcement
+
+Allowed write scope: none unless the orchestrator assigns exact, disjoint security test, policy, or release-review paths.
+
+Forbidden actions: do not edit shared state, stage, commit, push, merge, mutate providers, change credentials, submit builds, publish security terms, accept risks, or make founder-only decisions.
+
+## Required Handoff
+
+Return only these headings:
+
+- Scope reviewed
+- Evidence
+- Findings
+- Recommendations
+- Files changed
+- Validation
+- Risks and blockers
+- Proposed state patch
+- Knowledge receipt (`CONTRACT_FILES_LOADED`, `KNOWLEDGE_LOADED`, `ROLE_KNOWLEDGE_USED`, `SKILLS_USED`, `TOOLS_USED`; use `none` with a reason where a conditional route did not match)

package/examples/workspace/business/growth/DEMO_VIDEO.md

@@ -0,0 +1,44 @@
+# {{APP_NAME}} Demo Video Packet
+
+Status: partial
+Owner: design-guru + marketing-guru + orchestrator
+
+`growth/DEMO_VIDEO.md` records app-preview, launch-demo, support-demo, and social-demo proof. Use it when in-app iOS Simulator recording (Cmd+R), MobAI recorder, native iOS capture, desktop recorder, Remotion, Higgsfield reframe, or store App Preview work is in scope. Source footage must show the real app or explicitly approved prototype footage.
+
+## Source Inputs
+
+| Source | Role | Status |
+| --- | --- | --- |
+| `11_STAR_EXPERIENCE.md` | magical moment and line of feasibility | blocked |
+| `EMOTIONAL_DESIGN.md` | Emotional North Star and Experience Card for the hook | blocked |
+| `strategy/BRAND.md` | voice, owned words, banned claims | blocked |
+| `DESIGN.md` | tokens, type, motion, caption style | blocked |
+| `SCREENSHOTS.md` | App Preview and screenshot story | blocked |
+| In-app iOS Simulator (Cmd+R) / MobAI / native iOS / device capture | real in-app footage | blocked |
+| `CONTENT_ASSETS.md` + manifest | route, license, output registry | blocked |
+
+## Storyboard
+
+| Clip | Surface | First frame | 3-5s hook | Main proof | CTA/caption | Status |
+| --- | --- | --- | --- | --- | --- | --- |
+| App Preview 1 | App Store | Pending | magical V1 moment, muted-readable | real in-app result | Pending | blocked |
+| Launch demo | landing/social/support | Pending | user problem in plain language | real flow walkthrough | Pending | blocked |
+
+## Capture And Edit Ledger
+
+| Step | Tool/route | Input | Output | Proof | Status |
+| --- | --- | --- | --- | --- | --- |
+| Raw capture | in-app iOS Simulator Cmd+R (Desktop → `demo/raw/`), MobAI recorder, Codex Desktop native iOS/XcodeBuildMCP, serve-sim, or approved fallback | device/session | `demo/raw/` | command or recording log | blocked |
+| Edit/render | Remotion or approved editor | raw capture + captions | `demo/final/` | render command/log | blocked |
+| Reframe | Higgsfield reframe or local render variants | master clip | store/social aspect variants | output manifest | blocked |
+| QA | visual + claims review | final clips | approval notes | founder gate | blocked |
+
+## Truth And Accessibility Checks
+
+- [ ] First 3-5 seconds work muted and show the product payoff, not a logo/splash.
+- [ ] Captions are readable on mobile and do not cover critical UI.
+- [ ] No fake UI, unsupported claim, fabricated count, fake urgency, or bait-and-switch appears.
+- [ ] Footage recorded from an agent-driven simulator used a fixture or sandbox account; no real account, real user data, or production credential is visible in any frame.
+- [ ] Reduced-motion or low-motion alternative exists when animated effects carry meaning.
+- [ ] Output paths, license status, source inputs, and approval status are mirrored in `growth/content-assets/manifest.json`.
+- [ ] Founder approval is recorded before public posting, paid ads, store upload, or final App Preview submission.

package/examples/workspace/business/growth/EMAIL_OPS.md

@@ -0,0 +1,109 @@
+# Email Ops
+
+Status: partial until Resend domain, SPF, DKIM, DMARC, unsubscribe, and test-send evidence are recorded.
+
+---
+
+## Definition of Good
+
+### PRESENT — lane can be partial
+
+All three artifacts must exist before the email lane leaves draft:
+
+1. **Sender map** — every from-address, reply-to, subdomain, purpose, owner, and unsubscribe classification documented in the table below.
+2. **Sending domain** — the subdomain used for production sends recorded (e.g. `mail.example.com`). `resend.dev` is not acceptable for production.
+3. **Template inventory** — each email has a named template file or hosted-template reference in the table below.
+
+### PROVEN — hard errors when lane is done
+
+The following evidence paths must exist on disk and be non-empty before the email lane can be marked `done`. Missing paths block launch.
+
+| Proof artifact | Expected path (relative to repo root) | What it proves |
+| --- | --- | --- |
+| Domain verification screenshot or CLI output | `proof/email-domain-verified.*` | Resend dashboard shows the sending domain as Verified |
+| SPF + DKIM pass headers | `proof/email-spf-dkim-pass.*` | MX Toolbox / Gmail header check showing SPF=pass, DKIM=pass for the sending domain |
+| Successful test-send log | `proof/email-test-send-log.*` | At least one real transactional email delivered to an external inbox (Gmail, iCloud, or Outlook) recorded |
+| Keys via Doppler, not raw .env | `SECRETS.md` — must contain `RESEND_API_KEY` and reference Doppler | API keys routed through Doppler; raw `.env` with Resend keys is a secrets violation |
+
+Accepted artifact formats: `.png`, `.jpg`, `.json`, `.txt`, `.md`, or any non-empty file matching the glob. The check uses glob `proof/email-domain-verified*`, `proof/email-spf-dkim-pass*`, `proof/email-test-send-log*`.
+
+### OPTIMIZED — warnings (taste stays human)
+
+The following lifecycle automations are expected for a complete B2C launch. Each missing automation fires a warning; the founder can override by adding a `<!-- email-automation-override: <reason> -->` comment in this file or by marking the lane `not_needed`.
+
+| Automation | Trigger event | Status |
+| --- | --- | --- |
+| Trial expiry reminder | `trial.expiring` (24h before) | <!-- TODO: set to done/skipped --> |
+| Payment failure / billing recovery | `payment.failed` | <!-- TODO: set to done/skipped --> |
+| Welcome / activation | `user.created` or first value event | <!-- TODO: set to done/skipped --> |
+| Onboarding resume nudge | `onboarding.started` without `onboarding.completed` | <!-- TODO: set to done/skipped --> |
+| Cancellation confirmation + optional win-back | `subscription.cancelled` | <!-- TODO: set to done/skipped --> |
+
+Automation statuses are freeform. The validator scans for the string `done` or `skipped` next to each automation name. An automation row that still reads `TODO` triggers a warning.
+
+---
+
+## Brand Tokens
+
+Production templates pull their design system from `DESIGN.md` — never hardcoded hex values. `business/growth/resend/email-templates.ts` documents the contract (`LaunchEmailDesignSystem`: colors, typography, radius, spacing, email layout). Record the mapping here before production sends; `check:email` flags a missing mapping.
+
+| LaunchEmailDesignSystem field | DESIGN.md token | Value basis |
+| --- | --- | --- |
+| colors.accent / accentText | <!-- e.g. color.accent.primary --> | <!-- DESIGN.md section --> |
+| typography.bodyFontFamily | <!-- e.g. font.body --> | <!-- DESIGN.md section --> |
+| radius.card / radius.button | <!-- e.g. radius.md --> | <!-- DESIGN.md section --> |
+| spacing.* / email.maxWidth | <!-- e.g. space scale --> | <!-- DESIGN.md section --> |
+
+---
+
+## Sender Map
+
+| Email | Trigger | From address | Reply-to | Template | Unsubscribe required | Proof |
+| --- | --- | --- | --- | --- | --- | --- |
+| welcome | account creation or first value event | <!-- e.g. hello@mail.example.com --> | <!-- e.g. support@example.com --> | resend/email-templates.ts `welcomeEmail` | no (transactional) | operations/PROVIDER_PROOF.md |
+
+Add one row per email type. Mark unsubscribe as `yes (lifecycle)`, `yes (marketing)`, or `no (transactional)`.
+
+---
+
+## Domain / DNS Status
+
+| Field | Value |
+| --- | --- |
+| Sending domain/subdomain | <!-- e.g. mail.example.com --> |
+| Resend region | <!-- us-east-1 or eu-west-1 --> |
+| SPF record | <!-- TXT value or "pending" --> |
+| DKIM record | <!-- CNAME value or "pending" --> |
+| Custom return path | <!-- subdomain or "default" --> |
+| DMARC record | <!-- TXT value or "pending" --> |
+| Domain verification status | <!-- Verified / Pending --> |
+| Verification timestamp | <!-- ISO date --> |
+
+---
+
+## API Key Inventory
+
+| Key name | Permission | Domain scope | Env var | Deploy target | Rotation date | Owner |
+| --- | --- | --- | --- | --- | --- | --- |
+| production-transactional | sending_access | mail.example.com | RESEND_API_KEY | Doppler → Vercel/Workers | <!-- date --> | <!-- owner --> |
+
+API keys must be server-only (never `NEXT_PUBLIC_*` or mobile plist). Store via Doppler; record location in `SECRETS.md`.
+
+---
+
+## Validation Log
+
+Record each result here after running. All items must be checked before marking the lane `done`.
+
+- [ ] Domain/subdomain verified in Resend — evidence at `proof/email-domain-verified.*`
+- [ ] SPF and DKIM pass — evidence at `proof/email-spf-dkim-pass.*`
+- [ ] DMARC record exists and policy is justified
+- [ ] API key works in staging and production; no browser exposure
+- [ ] Test transactional email sent to Gmail/iCloud/Outlook — log at `proof/email-test-send-log.*`
+- [ ] Idempotency retry does not duplicate a transactional email
+- [ ] Webhook signature verification succeeds with raw body
+- [ ] Bounce/complaint path tested
+- [ ] Topic/unsubscribe behavior tested for lifecycle automations
+- [ ] Automation trigger and run status checked
+- [ ] `trust/PRIVACY.md` and `trust/TERMS.md` updated with Resend as email processor
+- [ ] `SECRETS.md` lists `RESEND_API_KEY` routed via Doppler

package/examples/workspace/business/growth/FASTLANE_OPS.md

@@ -0,0 +1,5 @@
+# Fastlane Ops
+
+Status: partial until winning formats, queue, channel rules, and public-post approval are recorded.
+
+Fastlane should only operationalize already-approved assets and claims. Founder approval is required before scheduling, posting, or paid distribution.

package/examples/workspace/business/growth/LAUNCH_NARRATIVE.md

@@ -0,0 +1,241 @@
+# Launch Narrative & Cadence — <App Name>
+
+> Generated by the B2C App Builder skill. Read alongside
+> [`knowledge/growth/launch-narrative-cadence.md`](../../../../knowledge/growth/launch-narrative-cadence.md), which
+> carries the full six-phase X launch checklist this artifact is built from.
+>
+> Operating premise: a launch is not the end of building the product — it is the reason to build it
+> a certain way. Use rare tentpole moments to earn a standing audience, then feed that audience with
+> a weekly feature cadence so it never goes cold. **The launch never ends.**
+
+Replace every `<...>` placeholder with the real decision. Keep all post copy in fenced code blocks
+and reviewed against the 2026 DO-NOT-DO list in `launch-narrative-cadence.md`. Keep public copy limited to
+what is true and attributable.
+
+---
+
+## Fit Gate
+
+- Lane status (`lanes.growth.status` in `state/business-state.json`): `<partial | done | not_needed | deferred>`
+- Is there a product moment heavy enough to back a tentpole? `<yes/no — what it is>`
+- If no tentpole yet: ship the weekly Feature Launch Cadence only, and record the tentpole as a future gate.
+
+## Launch Thesis (Feeling-First)
+
+Shape the launch around a feeling first, the feature second. If a reader can only react with
+"okay" and keep scrolling, the algorithm reads no reaction as no value.
+
+- Emotional North Star (from [`EMOTIONAL_DESIGN.md`](../product/experience/emotional-design/EMOTIONAL_DESIGN.md)): `<one line>`
+- The feeling this launch must create before the feature is explained: `<one line>`
+- Why this is in the zeitgeist this week: `<the trend/news/mood we are attaching to>`
+
+## Emotional Angle
+
+Pick the dominant angle and write to it. Strong launches usually land on one of:
+
+- **A glimmer of hope** — a hopeful take in a noisy, cynical category.
+- **Pointing at something broken** — name what is broken and say we are the ones fixing it.
+- **A category reframe** — a genuinely new take that changes how people think about the whole space.
+
+Chosen angle: `<hope | broken | reframe>` — `<one line on how this post expresses it>`
+
+---
+
+## Phase 1 — Foundation
+
+### Bold Claim
+
+One sentence: polarizing, defensible, novel. Test: would someone screenshot it to argue with a friend?
+
+- Bold claim: `<e.g. "See which deals close this quarter before they do." — not "AI-powered dashboard">`
+
+### Outcome Over Feature
+
+Features tell; outcomes sell. Write the outcome first, then map features to it.
+
+| Feature     | Outcome the user walks away with |
+| ----------- | -------------------------------- |
+| `<feature>` | `<outcome>`                      |
+
+### Network & Activatable Communities
+
+- Friendly, influential accounts to activate on day one (warm beats big): `<names + niche fit + are-their-followers-our-customers>`
+- Pre-written posts drafted for friendly accounts: `<where stored>`
+- Activatable communities (easy to excite / easy to anger): `<list + how each reacts to bold messaging>`
+
+### Marketing At The Table
+
+The roadmap is content. Sequence features so each release gives a fresh bold claim, and choose what
+ships next by what is interesting and trendy now — not just what is ready.
+
+- Roadmap review cadence with marketing in the room: `<e.g. monthly>`
+- Zeitgeist scan owner / where recorded: `<who / where>`
+- Next 3 releases sequenced as postable hooks: `<list>`
+
+## Phase 2 — Outlier Research
+
+Spend 10–20 hours before writing a word. You are not inventing content; you are finding what works
+and adapting proven formats.
+
+- 20+ outliers logged (500+ likes / 50+ comments / 50+ reposts ≈ 10x normal): `<link to research doc>`
+- Anatomy documented per outlier (hook, structure, emotional trigger, visual format, engagement pattern): `<where>`
+- Conversation map (what the niche complains about / celebrates / argues over): `<summary>`
+- 3–5 locked hook formats: `<list>`
+
+## Phase 3 — Content Creation
+
+- Every influencer post is written by us, word for word, with custom visuals — never a brief.
+- Format matched per creator (e.g. Ways I Use It / Comparison / Founder Story): `<mapping>`
+- Launch video built to the rules (bold claim in first 5s, present tense, sub-fifth-grade language,
+  audio/visual 1:1, viewer's outcome) — see [`CONTENT_ASSETS.md`](./content-assets/CONTENT_ASSETS.md): `<asset>`
+- One defensible rage-bait line (true, on-claim, subtle enough that most scroll past): `<line>`
+  - Ethics check: reviewed against the truthfulness/claims rule; this is a HIGH-sensitivity tactic. `<attestation>`
+
+## Phase 4 — Amplification Roster
+
+Founder-gated spend — route through [`growth/PAID_UA.md`](./PAID_UA.md).
+
+- Roster + budget (agency bands: ~$500–$3,000/influencer, ~$1,000 avg; 30–100 creators; $50K+ for a major launch): `<plan>`
+- Vetting rule: engagement quality, niche fit, followers who ARE our customer, creators who genuinely use tools like ours.
+- Confirmed: every creator will work their own comment section. `<yes/no>`
+
+---
+
+## Two Launch Types
+
+### Tentpole Launch (the "Super Bowl")
+
+Rare, planned weeks out; ~48-hour peak. Must feel special and be backed by a heavy product moment.
+
+- Trigger / product moment: `<what>` · Principal (a credible human, not the brand handle): `<who>`
+- Hero asset: `<video/clip>` · Lead time + owner: `<weeks out, who>`
+
+### Feature Launch Cadence (the heartbeat)
+
+Small, sharp, consistent releases packaged like events — the real way you stay in the conversation.
+
+- Cadence: `<e.g. one shippable, postable release per week>`
+- This quarter's candidate releases (sequenced by what is interesting): `<list>`
+
+### Algorithm Compounding
+
+The tentpole earns a **standing audience** (a slice of its reach keeps getting re-served for months),
+so weekly posts ride the wave instead of fighting a cold start.
+
+- Standing-audience signal watched: `<followers gained, profile-served impressions, returning viewers>`
+
+---
+
+## Tentpole Announcement Post
+
+Hook first, feeling first, feature second. No hashtags, no emojis, no link in this post.
+
+```text
+<Hook line that makes someone feel something.>
+
+<One or two lines naming the broken thing or the hopeful reframe.>
+
+Introducing <App Name> — <the one-line what-it-is>.
+
+<Video below.>
+```
+
+## Copy Guardrails
+
+The 2026 do-not-do list (checked deterministically against the fenced post copy):
+
+- No hashtags. No emojis carrying the message. No link in the main post — link goes in the first reply.
+- No generic "we shipped X" copy that earns an "okay" and a scroll. Make it lovable or hateable.
+- Lead with the hook/POV, not a preamble. No nuance in the post — save nuance for the replies.
+- No "congrats!" seeded replies — every pre-briefed reply adds a distinct angle.
+
+## Feature Launch Post Template
+
+Reusable weekly beat. Same rules: hook first, no hashtags, no emojis, link in the reply.
+
+```text
+<Sharp one-line hook tied to a real pain or a delightful result.>
+
+Launching today: <the feature in plain words>.
+
+<Short proof line — what it does for you.>
+
+<Video below.>
+```
+
+---
+
+## Phase 5 — Launch Day Run-of-Show
+
+Plan every beat weeks out — down to post order and who replies to what.
+
+| Time (relative) | Beat                                            | Owner       | Asset      | Notes                                      |
+| --------------- | ----------------------------------------------- | ----------- | ---------- | ------------------------------------------ |
+| T-0             | Principal posts the announcement (hook + video) | `<founder>` | hero video | link goes in the FIRST reply, not the post |
+| T+2m            | Principal reply with the link + how to try it   | `<founder>` | reply      |                                            |
+| T+5m            | Seeded allies quote/reply with their angle      | `<names>`   | replies    | pre-briefed, staggered                     |
+| T+1h            | Second-wave amplifiers / partners               | `<names>`   | replies    |                                            |
+| T+1d            | Recap / "what people are saying" follow-up      | `<founder>` | follow-up  | feeds the long tail                        |
+
+### Launch-Day Timing Engine
+
+One principle: make it feel organic — layer amplification gradually, starting closest to your ICP.
+
+- **Gate 1 — Let it breathe (0–60 min):** post live, do nothing externally, reply to every comment,
+  watch likes/min, reply quality, retweet pace. Strong velocity → proceed; weak → fix before amplifying.
+- **Gate 2 — First wave (hours 1–2):** deploy 10–20 influencers staggered in 30-min windows, biggest
+  ICP first; each retweets the main post, runs its own giveaway, and works its comments.
+- **Gate 3 — Full deployment + live adjustment (hours 2–3):** add theme pages / Discord where momentum
+  builds, hold back where it slips, coordinate cross-engagement.
+- **Reply to everything for 48 hours** — author-reply chains are the strongest signal; substantive replies, not "thanks".
+
+## Phase 6 — Follow-Up Window
+
+The momentum window after a viral post boosts what you post inside it. Don't waste it.
+
+- Day 2: meme or short AI-generated video riffing on the launch.
+- Day 3: deep dive on one feature.
+- Day 4: founder story / behind-the-scenes.
+- Week 2: results, testimonials, early traction.
+
+## Sourcing And Ranking
+
+The platform's two signals (the agency's working model, not official docs), used as a planning lens:
+
+- **Sourcing — will it be shown?** Retweets qualify the post for the For You feed. This is why every giveaway requires a retweet.
+- **Ranking — how high?** Reply chains where you respond back; conversation = time on site. Every author reply creates a new chain.
+
+---
+
+## Measurement Plan
+
+Tie attention to product behavior — see [`analytics/ANALYTICS.md`](../analytics/ANALYTICS.md) and
+[`analytics/analytics-plan.html`](../analytics/analytics-plan.html).
+
+- Funnel watched: post impressions → reply-link clicks (UTM'd) → install → activation event `<name>`
+- Self-reported attribution source key for this channel (per `analytics/ANALYTICS.md` attribution contract): `<x_twitter | creator | ...>`
+- Standing-audience metric reviewed weekly: `<metric>`
+- Target bands (agency-stated, aim-at not promised): ~2–10M views, 500–2,000+ comments (all replied to), 1,000–5,000+ retweets.
+
+## Stop And Scale Rules
+
+- Scale a format when: `<e.g. it clears the activation bar two weeks running>`
+- Kill or rework a format when: `<e.g. no reaction = no value; below baseline twice>`
+- Save tentpoles for genuinely heavy moments; do not cry "Super Bowl" every week.
+
+## Founder-Only Gates
+
+- Public posting/scheduling and the launch go-live decision.
+- Connecting or posting from brand/founder social accounts.
+- Any paid amplification or creator spend (route through [`growth/PAID_UA.md`](./PAID_UA.md)).
+- Unverifiable claims (view counts, revenue, "fastest-growing") — keep public copy to what is true.
+
+## Traceability
+
+- Emotional North Star and feeling: [`EMOTIONAL_DESIGN.md`](../product/experience/emotional-design/EMOTIONAL_DESIGN.md)
+- Magical V1 moment the launch sells: [`11_STAR_EXPERIENCE.md`](../product/experience/11-star-experience/11_STAR_EXPERIENCE.md)
+- Hero video and post assets: [`CONTENT_ASSETS.md`](./content-assets/CONTENT_ASSETS.md)
+- Product-led sharing/referral loop this rides: [`VIRAL_GROWTH.md`](./VIRAL_GROWTH.md)
+- Store launch calendar and listing alignment: [`APP_STORE_LISTING.md`](../store/app-store-listing/APP_STORE_LISTING.md)
+- Events, attribution, dashboards: [`analytics/ANALYTICS.md`](../analytics/ANALYTICS.md)
+- Cross-surface launch decisions: [`state/LAUNCH_TRACE.md`](../state/LAUNCH_TRACE.md)