b2c-app-builder 0.221.39

package/agents/skills/b2c-app-builder/references/composition.md

@@ -0,0 +1,7 @@
+# Change packages or providers
+
+## Customize composition
+
+Discover capabilities, providers, and recipes with `b2c_discover` or `b2c catalog --json`. Preview a `b2c/v1` composition with `b2c_compose` or `b2c compose --config b2c.yaml --json`. Both share one versioned result contract. Read `canApply`, blockers, configuration, authority, and verification as separate facts. This declaration preview does not activate a workspace. For an installed runtime, import a local extension with `b2c package-import --workspace ID --source PATH`, inspect `b2c_packages`, then use `b2c_composition_plan` with exact package digests. Apply the returned preview through `b2c composition-activate`; local writes stay CLI-only. Fresh bootstrap activates the complete-business default after canonical product acceptance. Change an installed recipe only through composition activation. Imported package declarations do not install an execution route or prove provider readiness.
+
+`contracts/public-api/REFERENCE.md` holds the supported schema. Use this consumer contract across internal changes. Load this section when the task is changing packages or providers. It is not a prerequisite for creating or resuming a business.

package/agents/skills/b2c-app-builder/references/mobile-operation.md

@@ -0,0 +1,5 @@
+# Operate a mobile app
+
+## Mobile app operation
+
+For app exploration, flow walks, functional or design checks, screenshots, and recordings, discover `b2c/mobile-app-operation`. Honor explicit provider bindings. Otherwise prefer native tools the current host already exposes when they cover the target and operations. Choose MobAI or another provider for uncovered requirements. Never infer availability from the agent name or shrink the evidence scope. Use the existing native Route Ladder for compatible execution. Raw captures feed independent review or marketing composition. They do not prove acceptance on their own. The public mobile-capture recipe previews selection only. It cannot run a device.

package/agents/skills/b2c-app-builder/references/setup.md

@@ -0,0 +1,18 @@
+# Connect the builder
+
+Open this reference only for requested setup or connectivity needed by current managed execution. Focused advice does not require a connection.
+
+## Capability and compatibility
+
+Use the local `b2c-local` MCP server from this package for workspace planning and execution. Hosted knowledge registers as `b2c-hosted` and cannot see or run a local business. When both are configured, select by that capability; a leftover `b2c-app-builder` name is not a third surface. Duplicate names are a collision. A missing worker CLI degrades local execution health; it does not turn this connection into hosted knowledge. Degraded execution still selects b2c-local. Leftover CLI-only public MCP names stay CLI-only on this local connection. Leftover write-gated MCP names stay CLI-only on this local connection when writes are mcp_readonly. Hosted leftover names stay wrong-surface. Do not route consumer-app work through Planes.
+
+## Connect when authorized
+
+This is a connectivity gate, not the business start path. If the local MCP is unavailable:
+
+1. Run `b2c inspect` when the CLI exists. `b2c doctor` is a supported equivalent.
+2. When the user asked for setup, install this package's dependencies, link the package, and run `b2c setup`.
+3. Use the exact MCP registration command that setup prints.
+4. Keep the MCP read-only by default. Use the CLI for approved writes.
+
+Do not edit an agent configuration or install software unless the user requested setup. For managed work, use the [business lifecycle](business-lifecycle.md): create or resume, then status, then plan.

package/agents/skills/b2c-define-product/SKILL.md

@@ -0,0 +1,49 @@
+---
+name: b2c-define-product
+description: "Define or review a consumer product from supplied evidence and accepted constraints. Preserve complete scope without restarting opportunity research or scaffolding an app."
+compatibility: Markdown and supplied evidence. Provider-backed work needs the corresponding authorized tools. The B2C runtime is optional for focused advisory work.
+metadata:
+  source-workflow: "workflow.research.research-backed-spec"
+  generated-by: b2c-catalog
+---
+
+# Define or refine a product
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+## Choose the scope
+
+Use this task directly for focused advice, review, or an authorized change. Do not require setup, install software, create a workspace, or activate a full launch graph merely to use this expertise. A review is read-only unless the user also requests changes.
+
+For an existing managed business, read business-status then business-plan and use its current brief. This skill cannot choose executable next work, bypass prerequisites, change provider bindings, or replace runtime acceptance. Do not inspect raw reducer files merely to start a focused task.
+
+## Method
+
+For a review, assess the existing evidence against this method and return findings; do not execute its authoring or mutation instructions. For requested creation or implementation, follow the method only within the accepted scope and authority.
+
+## Continuous experience principle
+
+Carry the accepted user's intended outcome through the work: make the result meaningfully better for that user, preserve relevant identity, hierarchy, accessibility, and recovery constraints, and verify behavior as well as presence. Apply this only where relevant; a narrow fix stays narrow. This principle does not require an 11-star exercise or a numeric taste score.
+
+Identify the user, the recurring problem, the useful outcome, the repeated action, and the
+moment a user can see why this product is better than the alternatives. Select a platform
+because it serves that loop, not because a starter happens to exist.
+
+Use decisions delegated in the opening mandate. Do not ask the founder to repeat an already
+supplied choice. Escalate a genuinely ambiguous product decision only when its authority was
+not delegated. Record product meaning in `product.yaml`; render `PRODUCT.md`. Never write
+product decisions into reducer-owned execution state.
+
+## Load only what the task needs
+
+[Task inputs, outputs, checks, and knowledge selectors](references/task.md) supplies the canonical details when they are needed. Open the specific referenced sections, not the whole library. In connected knowledge retrieval, follow exact section selectors, revision hashes, and continuation calls. Unresolved guidance remains unresolved.
+
+## Tools and evidence
+
+Keep the method independent of the agent host and business provider. Honor explicit selections. Use already available, authorized tools that implement the required operation; do not infer availability from the agent's name or silently substitute a provider. Load a provider's procedure only when its action is current. Record missing capabilities without inventing provider proof.
+
+Pause for access or secrets, spend, pricing or legal decisions, destructive changes, hosted deployment, store submission, or production release. Guidance is not permission. A proposed price is not an approved price, and a mock is not live evidence.
+
+## Return
+
+Report findings or changes, the evidence inspected, applicable checks actually run, unresolved requirements, and the next decision. Label advisory findings separately from accepted business evidence. Do not record business completion or modify reducer-owned state outside the supported runtime.

package/agents/skills/b2c-define-product/references/task.md

@@ -0,0 +1,42 @@
+# Consumer app opportunity and product research
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.research.research-backed-spec`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product.yaml`
+
+Consult when relevant: None declared.
+
+Managed outputs: `strategy/RESEARCH.md`, `strategy/SIGNAL_CORPUS.md`, `strategy/OFFER_TEST.md`, `product.yaml`, `PRODUCT.md`
+
+## Verification and authority
+
+Declared checks: `check:research-workflow-output`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Consumer Craft Benchmarks](../../../../knowledge/design/consumer-craft-benchmarks.md) | calibrating complete consumer-business mobile and landing craft against primary-source product examples before production and independent review | `reference.design.consumer-craft-benchmarks` |
+| [Design Worthiness](../../../../knowledge/design/design-worthiness.md) | when DESIGN.md authoring or its generated Design Room review needs the mechanical worthiness floor or its independent audit must record the delegated taste decision before engineering harden, including native flow semantics and mechanical anti-generic consistency, with no beauty score | `reference.design.design-worthiness` |
+| [iOS Consumer Playbook Intake](../../../../knowledge/growth/ios-consumer-playbook-intake.md) | before encoding an external iOS consumer-app playbook into paywall, onboarding, or paid-UA gates, including ToS-risky spend and social-proof tactics | `reference.growth.ios-consumer-playbook-intake` |
+| [Offer And Pricing Decisions](../../../../knowledge/money/offer-and-pricing-decisions.md) | research proposes an offer or monetization locks a production price; separate hypotheses, offer evidence, and exact founder authority | `reference.money.offer-and-pricing-decisions` |
+| [Artifact Contracts](../../../../knowledge/process/artifact-contracts.md) | writing docs, handoff bundles, or acceptance criteria | `reference.process.artifact-contracts` |
+| [Research Intelligence](../../../../knowledge/process/tool-recipes/research-intelligence.md) | delegating AppKittie store intelligence, XPOZ social-language, last30days recency briefs, Firecrawl web intelligence, Refero UX pattern research, or name/keyword collision checks — routed from tool-recipes.md; complements the Market research rows | `reference.process.tool-recipes.research-intelligence` |
+| [Core Loop And Complete Product Scope](../../../../knowledge/product/core-loop-and-complete-scope.md) | the product does not match one of the four shipped app archetypes (habit tracker, photo/AI media, social network, AI chat companion) — before naming the core loop, defining every required system in the accepted product, or specing schema for any other product shape | `reference.product.core-loop-and-complete-scope` |
+| [Product Moat](../../../../knowledge/product/product-moat.md) | before the spec or launch surfaces harden — the moat and defensibility test for what makes this app hard to copy; see eleven-star-experience.md for the 11-Star Run Protocol itself | `reference.product.product-moat` |
+| [Distribution-First Niche](../../../../knowledge/research/distribution-first-niche.md) | before Phase 2 build; when research must name a paying audience, one reachable channel, and purchases as validation | `reference.research.distribution-first-niche` |
+| [Distribution, Offer, And Signal Validation](../../../../knowledge/research/distribution-offer-signal.md) | before a Go, Pivot, or Kill decision; when founder posts, transcripts, reviews, interviews, support records, or prior launch material can inform the business; before automating a value-producing business loop | `reference.research.distribution-offer-signal` |
+| [Go, Pivot, Or Kill: The Pre-Build Research Verdict](../../../../knowledge/research/go-pivot-or-kill.md) | writing or updating strategy/RESEARCH.md's Category Revenue Reality or Go, Pivot, Or Kill sections; deciding whether a category and wedge earn a build before Phase 2 design/build spend | `reference.research.go-pivot-or-kill` |
+| [Transformation Demo](../../../../knowledge/research/transformation-demo.md) | before a consumer app Go verdict; when strategy/RESEARCH.md needs a one-screenshot 15s transformation demo that is not a vitamin feature | `reference.research.transformation-demo` |

package/agents/skills/b2c-design-onboarding/SKILL.md

@@ -0,0 +1,51 @@
+---
+name: b2c-design-onboarding
+description: "Design, review, or improve a consumer app's onboarding and first-value journey. Classify a focused audit, incremental change, or full redesign before selecting the relevant research, flow, state, accessibility, and verification guidance. Do not start a full rebuild for a small signup fix."
+compatibility: Markdown and supplied evidence. Provider-backed work needs the corresponding authorized tools. The B2C runtime is optional for focused advisory work.
+metadata:
+  source-workflow: "workflow.experience.onboarding-system.onb-00-resume-scope"
+  generated-by: b2c-catalog
+---
+
+# Design or review onboarding
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+## Choose the scope
+
+Use this task directly for focused advice, review, or an authorized change. Do not require setup, install software, create a workspace, or activate a full launch graph merely to use this expertise. A review is read-only unless the user also requests changes.
+
+For an existing managed business, read business-status then business-plan and use its current brief. This skill cannot choose executable next work, bypass prerequisites, change provider bindings, or replace runtime acceptance. Do not inspect raw reducer files merely to start a focused task.
+
+## Method
+
+For a review, assess the existing evidence against this method and return findings; do not execute its authoring or mutation instructions. For requested creation or implementation, follow the method only within the accepted scope and authority.
+
+## Continuous experience principle
+
+Carry the accepted user's intended outcome through the work: make the result meaningfully better for that user, preserve relevant identity, hierarchy, accessibility, and recovery constraints, and verify behavior as well as presence. Apply this only where relevant; a narrow fix stays narrow. This principle does not require an 11-star exercise or a numeric taste score.
+
+| Mode          | Use when                                             | Cutover rule                             |
+| ------------- | ---------------------------------------------------- | ---------------------------------------- |
+| `greenfield`  | No production onboarding exists                      | Build only the target system             |
+| `replacement` | Existing onboarding is rebuilt from first principles | Hard cutover; no permanent coexistence   |
+| `audit_only`  | Findings are requested without implementation        | Produce evidence, target graph, and plan |
+| `incremental` | The founder explicitly limits scope                  | Preserve only the named boundary         |
+
+Rebuild, replace, standardize, and rethink requests default to `replacement`. Preserve durable user value through an isolated, rehearsed, one-time transformation, then delete the transformation and every obsolete route, state, event, provider object, test, and document.
+
+## Load only what the task needs
+
+[Task inputs, outputs, checks, and knowledge selectors](references/task.md) supplies the canonical details when they are needed. Open the specific referenced sections, not the whole library. In connected knowledge retrieval, follow exact section selectors, revision hashes, and continuation calls. Unresolved guidance remains unresolved.
+
+For onboarding beyond scope classification, choose the relevant [task stage](references/stages.md). An audit or small change does not imply a full rebuild; a complete redesign preserves all applicable catalog obligations.
+
+## Tools and evidence
+
+Keep the method independent of the agent host and business provider. Honor explicit selections. Use already available, authorized tools that implement the required operation; do not infer availability from the agent's name or silently substitute a provider. Load a provider's procedure only when its action is current. Record missing capabilities without inventing provider proof.
+
+Pause for access or secrets, spend, pricing or legal decisions, destructive changes, hosted deployment, store submission, or production release. Guidance is not permission. A proposed price is not an approved price, and a mock is not live evidence.
+
+## Return
+
+Report findings or changes, the evidence inspected, applicable checks actually run, unresolved requirements, and the next decision. Label advisory findings separately from accepted business evidence. Do not record business completion or modify reducer-owned state outside the supported runtime.

package/agents/skills/b2c-design-onboarding/references/stages.md

@@ -0,0 +1,31 @@
+# Design or review onboarding: task stages
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Choose only the stage that matches the requested change or current business brief. This index preserves the catalog's internal work units; it is not an instruction to execute every stage. A complete managed redesign uses the existing planner and its gates, not this table as a substitute scheduler.
+
+| Stage | Relevant task |
+| --- | --- |
+| [Onboarding ONB-00: resume and classify scope](task.md) | Consumer onboarding work begins or resumes; classify greenfield, replacement, audit-only, or a bounded incremental onboarding change |
+| [Onboarding ONB-01: current-state trace](workflow.experience.onboarding-system.onb-01-current-state-trace.md) | Trace the actual onboarding implementation, documents, routes, state, providers, events, failures, tests, and replaced surfaces. Founder phrasing: signup behaves like this today, screen by screen. |
+| [Onboarding ONB-02: evidence plan](workflow.experience.onboarding-system.onb-02-evidence-plan.md) | Define the onboarding evidence hierarchy, access constraints, sample plan, and freshness cutoff |
+| [Onboarding ONB-03: current guidance research](workflow.experience.onboarding-system.onb-03-current-guidance.md) | Research current consumer onboarding evidence, platform guidance, benchmarks, and practitioner heuristics |
+| [Onboarding ONB-04: competitor review analysis](workflow.experience.onboarding-system.onb-04-competitor-reviews.md) | Mine direct and adjacent competitor reviews, including negative themes and a positive-review control |
+| [Onboarding ONB-05: authorized flow atlas](workflow.experience.onboarding-system.onb-05-onbo-hub-atlas.md) | Build an authorized Onbo Hub flow atlas without scraping, bypassing access controls, or inferring locked screens |
+| [Onboarding ONB-06: internal guidance audit](workflow.experience.onboarding-system.onb-06-internal-guidance-audit.md) | Audit applicable B2C App Builder and internal B2C guidance and resolve conflicts or outdated rules |
+| [Onboarding ONB-07: provider and policy landscape](workflow.experience.onboarding-system.onb-07-provider-policy-landscape.md) | Refresh monetization, identity, analytics, RevenueCat, billing, platform-policy, and regional capability facts |
+| [Onboarding ONB-08: motion research](workflow.experience.onboarding-system.onb-08-motion-research.md) | Research interaction and motion for the selected surfaces and translate applicable references into the target framework |
+| [Onboarding ONB-09: evidence join](workflow.experience.onboarding-system.onb-09-evidence-join.md) | Join guidance, reviews, flow evidence, internal doctrine, provider facts, policy, and motion research into explicit decisions |
+| [Onboarding ONB-10: first value and activation](workflow.experience.onboarding-system.onb-10-first-value-activation.md) | Define first value rendered, first value engaged, activation, habit, and retention hypotheses |
+| [Onboarding ONB-11: effort and question audit](workflow.experience.onboarding-system.onb-11-effort-question-audit.md) | Audit effort before value, question usefulness, permissions, interruption budget, and visible personalization proof |
+| [Onboarding ONB-12: state and identity contract](workflow.experience.onboarding-system.onb-12-state-identity-contract.md) | Define canonical journey, profile, identity, entitlement, continuity, and cross-surface state transitions |
+| [Onboarding ONB-13: analytics and experiments](workflow.experience.onboarding-system.onb-13-analytics-experiments.md) | Define typed analytics, authoritative emitters, identity stitching, deduplication, experiment assignment, exposure, and expected sequences |
+| [Onboarding ONB-14: trust, lifecycle, and policy](workflow.experience.onboarding-system.onb-14-trust-lifecycle-policy.md) | Define review timing, permissions, lifecycle, privacy, security, accessibility, and policy behavior |
+| [Onboarding ONB-15: architecture decision](workflow.experience.onboarding-system.onb-15-architecture-decision.md) | Compare native-first, hosted-funnel-first, hybrid, web-first, and evidence-backed alternatives and select one target architecture |
+| [Onboarding ONB-16: canonical journey graph](workflow.experience.onboarding-system.onb-16-journey-graph.md) | Define acquisition-specific journeys that converge on one semantic onboarding state graph |
+| [Onboarding ONB-17: screen, control, and paywall contract](workflow.experience.onboarding-system.onb-17-screen-control-paywall-contract.md) | Specify every onboarding screen, copy key, control, action, paywall state, failure, recovery, accessibility, and localization behavior |
+| [Onboarding ONB-18: visual design and prototype](workflow.experience.onboarding-system.onb-18-visual-design-prototype.md) | Produce actual high-fidelity onboarding design, motion, an interactive prototype, and design QA |
+| [Onboarding ONB-19: implementation and cutover contract](workflow.experience.onboarding-system.onb-19-implementation-cutover-contract.md) | Define implementation units, reliability, performance, observability, privacy, migration, hard cutover, and deletion work |
+| [Onboarding ONB-20: adversarial QA](workflow.experience.onboarding-system.onb-20-adversarial-qa.md) | Run the synthetic one-star pre-mortem, policy review, instrumentation QA, accessibility review, and provider-realism review |
+| [Onboarding ONB-21: Compound Engineering plan](workflow.experience.onboarding-system.onb-21-compound-engineering-plan.md) | Translate the accepted onboarding graph into an implementation-ready Compound Engineering plan, or the B2C App Builder-equivalent fallback |
+| [Onboarding ONB-22: execute, cut over, and verify](workflow.experience.onboarding-conversion.md) | Implement or finalize the accepted onboarding graph, verify the canonical artifacts, cut over, and prove the target is the only runtime |

package/agents/skills/b2c-design-onboarding/references/task.md

@@ -0,0 +1,31 @@
+# Onboarding ONB-00: resume and classify scope
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-system.onb-00-resume-scope`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `analytics/ANALYTICS.md`, `product/experience/11-star-experience/11_STAR_EXPERIENCE.md`
+
+Consult when relevant: None declared.
+
+Managed outputs: `product/onboarding/graph/ONB-00-resume-scope.md`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-evidence-onb-00`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |

package/agents/skills/b2c-design-onboarding/references/workflow.experience.onboarding-conversion.md

@@ -0,0 +1,43 @@
+# Onboarding ONB-22: execute, cut over, and verify
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-conversion`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Method
+
+Execute ONB-21's Compound Engineering task list for real (implementation, review, tests, and provider validation), then assemble the canonical product/ONBOARDING.md by transcribing every ONB-00 through ONB-21 packet into its matching section, and mark every ONB-00 through ONB-22 row in the Graph Run table done with exactly one row per node. Only after the founder approves the hard cutover, delete the replaced runtime and transformation tooling per ONB-19's Removal Inventory, and mark a row's Disposition delete only once that artifact is actually gone from the repository — check:onboarding-cutover-repository verifies every claimed deletion against real filesystem state, not prose. Confirm RevenueCat and PostHog readiness against operations/PROVIDER_PROOF.md before claiming provider rows ready, produce product/onboarding/runtime-evidence.json for every applicable ONB-13 scenario with current schema, initialization, emitter and test source hashes, actual event sequences, collector readback and execution evidence, then re-render product/onboarding.html to match. Local trace fixtures never prove live delivery. check:onboarding-graph-complete, check:onboarding-page-fresh, check:onboarding-cutover-repository, and check:provider-proof-onboarding all gate this node, and none of them can be satisfied by prose alone.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product/onboarding/graph/ONB-00-resume-scope.md`, `product/onboarding/graph/ONB-01-current-state-trace.md`, `product/onboarding/graph/ONB-02-evidence-plan.md`, `product/onboarding/graph/ONB-03-current-guidance.md`, `product/onboarding/graph/ONB-04-competitor-reviews.md`, `product/onboarding/graph/ONB-05-onbo-hub-atlas.md`, `product/onboarding/graph/ONB-06-internal-guidance-audit.md`, `product/onboarding/graph/ONB-07-provider-policy-landscape.md`, `product/onboarding/graph/ONB-08-motion-research.md`, `product/onboarding/graph/ONB-09-evidence-join.md`, `product/onboarding/graph/ONB-10-first-value-activation.md`, `product/onboarding/graph/ONB-11-effort-question-audit.md`, `product/onboarding/graph/ONB-12-state-identity-contract.md`, `product/onboarding/graph/ONB-13-analytics-experiments.md`, `product/onboarding/graph/ONB-14-trust-lifecycle-policy.md`, `product/onboarding/graph/ONB-15-architecture-decision.md`, `product/onboarding/graph/ONB-16-journey-graph.md`, `product/onboarding/graph/ONB-17-screen-control-paywall-contract.md`, `product/onboarding/graph/ONB-18-visual-design-prototype.md`, `product/onboarding/graph/ONB-19-implementation-cutover-contract.md`, `product/onboarding/graph/ONB-20-adversarial-qa.md`, `product/onboarding/graph/ONB-21-compound-engineering-plan.md`, `product/copy/COPY_DECK.md`, `operations/PROVIDER_PROOF.md`
+
+Consult when relevant: None declared.
+
+Managed outputs: `product/ONBOARDING.md`, `product/onboarding.html`, `product/onboarding/runtime-evidence.json`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-graph-complete`, `check:onboarding-page-fresh`, `check:onboarding-cutover-repository-complete`, `check:provider-proof-onboarding`, `check:onboarding-foundations-runtime`
+
+Additional founder decisions: approve the hard cutover and replaced runtime deletion
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Analytics And Attribution](../../../../knowledge/data/analytics-attribution.md) | before onboarding, paywalls, funnels, store CTAs, referrals, lifecycle email, UGC/Fastlane campaigns, paid UA, or any builder prompt that names events; before PostHog setup, dashboards, deep links, feature flags, experiments, or session replay | `reference.data.analytics-attribution` |
+| [Commitment Funnel](../../../../knowledge/experience/commitment-funnel.md) | before first-session screen order; when product.yaml records feature.commitment-funnel as required and onboarding needs welcome, quiz, micro-commitment, personalized insight, hard paywall, and install-to-trial quality metrics | `reference.experience.commitment-funnel` |
+| [Eleven-Star Experience](../../../../knowledge/experience/eleven-star-experience.md) | before PRODUCT.md, DESIGN.md, onboarding, ads, store screenshots, content assets, or engineering plans are treated as ready; on "11-star run"/"11-star pass" — follow the reference's 11-Star Run Protocol before any other output | `reference.experience.eleven-star-experience` |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |
+| [Research-backed onboarding with identity and measurement](../../../../knowledge/experience/onboarding-foundations.md) | Before onboarding research acceptance, identity and analytics decisions, design lock, prototype acceptance or final onboarding verification. | `reference.experience.onboarding-foundations` |
+| [Paywall Goal Headline](../../../../knowledge/experience/paywall-goal-headline.md) | before paywall copy lock; when product.yaml records feature.paywall-goal-headline as required and the declared present-paywall owner can bind RevenueCat offering metadata | `reference.experience.paywall-goal-headline` |
+| [Push Notification Lifecycle](../../../../knowledge/experience/push-notification-lifecycle.md) | before push permission priming, opt-in timing, or notification lifecycle design; for transactional and lifecycle email mechanics load operations/resend-email-ops.md instead | `reference.experience.push-notification-lifecycle` |
+| [Consumer Copy Benchmarks](../../../../knowledge/words/consumer-copy-benchmarks.md) | when any surface a user reads is authored or reviewed — landing, store listing, screenshot captions, paywall, onboarding, lifecycle email, share text — and when a new app's voice is established in COPY_BRIEF.md / BRAND.md (the Voice benchmarks section writes from this file's method) | `reference.words.consumer-copy-benchmarks` |
+| [Conversion Copy](../../../../knowledge/words/conversion-copy.md) | before writing any words a user reads: conversion copy (hero/CTA, store, paywall) and every in-app string (buttons, empty states, errors, settings); and l10n readiness | `reference.words.conversion-copy` |

package/agents/skills/b2c-design-onboarding/references/workflow.experience.onboarding-system.onb-01-current-state-trace.md

@@ -0,0 +1,35 @@
+# Onboarding ONB-01: current-state trace
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-system.onb-01-current-state-trace`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Method
+
+Trace the onboarding implementation as it exists today, surface by surface: for each route, state slice, provider call, event, permission, paywall, and replaced code path, record owner, source of truth, persisted state, API/event contract, consumers, and failure behavior. Write the trace to product/onboarding/graph/ONB-01-current-state-trace.md — check:onboarding-evidence-onb-01 rejects a thin or placeholder packet, and ONB-02's evidence plan depends on knowing what actually exists before scoping new evidence collection.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product/onboarding/graph/ONB-00-resume-scope.md`
+
+Consult when relevant: None declared.
+
+Managed outputs: `product/onboarding/graph/ONB-01-current-state-trace.md`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-evidence-onb-01`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |

package/agents/skills/b2c-design-onboarding/references/workflow.experience.onboarding-system.onb-02-evidence-plan.md

@@ -0,0 +1,35 @@
+# Onboarding ONB-02: evidence plan
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-system.onb-02-evidence-plan`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Method
+
+Define the evidence hierarchy (rule, evidence, benchmark, observation, heuristic, hypothesis, or open question), the access constraints per source (e.g. authorized-only Onbo Hub access, no scraping), a sample plan per evidence node, and a freshness cutoff date beyond which a finding must be re-verified. Write the plan to product/onboarding/graph/ONB-02-evidence-plan.md so ONB-03 through ONB-08 collect against one shared standard instead of six competing ones; check:onboarding-evidence-onb-02 rejects a packet with no real plan.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product/onboarding/graph/ONB-01-current-state-trace.md`
+
+Consult when relevant: None declared.
+
+Managed outputs: `product/onboarding/graph/ONB-02-evidence-plan.md`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-evidence-onb-02`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |

package/agents/skills/b2c-design-onboarding/references/workflow.experience.onboarding-system.onb-03-current-guidance.md

@@ -0,0 +1,35 @@
+# Onboarding ONB-03: current guidance research
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-system.onb-03-current-guidance`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Method
+
+Research current consumer-onboarding evidence — platform guidance, published benchmarks, and practitioner heuristics — that bears on this product's onboarding decisions, each dated and sourced per ONB-02's evidence hierarchy and freshness cutoff. Write findings as Evidence Ledger rows in product/onboarding/graph/ONB-03-current-guidance.md; check:onboarding-evidence-onb-03 rejects a packet with no prose finding or a templated stub, and ONB-09 joins this evidence into explicit product decisions.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product/onboarding/graph/ONB-02-evidence-plan.md`
+
+Consult when relevant: None declared.
+
+Managed outputs: `product/onboarding/graph/ONB-03-current-guidance.md`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-evidence-onb-03`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |

package/agents/skills/b2c-design-onboarding/references/workflow.experience.onboarding-system.onb-04-competitor-reviews.md

@@ -0,0 +1,35 @@
+# Onboarding ONB-04: competitor review analysis
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-system.onb-04-competitor-reviews`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Method
+
+Mine direct and adjacent competitors' reviews for negative onboarding themes, each backed by dates, versions, and sample size, and pair every theme with a positive-review control so the finding isn't cherry-picked. Record each as a Competitor Review Matrix row (root-cause class, disposition, test) in product/onboarding/graph/ONB-04-competitor-reviews.md — report frequency only within the sample, and do not let an onboarding fix paper over an underlying product defect; check:onboarding-evidence-onb-04 rejects a stub packet.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product/onboarding/graph/ONB-02-evidence-plan.md`
+
+Consult when relevant: None declared.
+
+Managed outputs: `product/onboarding/graph/ONB-04-competitor-reviews.md`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-evidence-onb-04`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |

package/agents/skills/b2c-design-onboarding/references/workflow.experience.onboarding-system.onb-05-onbo-hub-atlas.md

@@ -0,0 +1,35 @@
+# Onboarding ONB-05: authorized flow atlas
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-system.onb-05-onbo-hub-atlas`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Method
+
+Build the authorized Onbo Hub flow atlas using only authorized access — do not scrape, bypass access controls, reuse credentials, or infer locked screens — recording screens, effort, first-value class, paywall placement, and review tension per flow, with an adopt/test/reject/investigate decision for each. Write the atlas to product/onboarding/graph/ONB-05-onbo-hub-atlas.md; check:onboarding-evidence-onb-05 rejects a stub or templated packet.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product/onboarding/graph/ONB-02-evidence-plan.md`
+
+Consult when relevant: None declared.
+
+Managed outputs: `product/onboarding/graph/ONB-05-onbo-hub-atlas.md`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-evidence-onb-05`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |

package/agents/skills/b2c-design-onboarding/references/workflow.experience.onboarding-system.onb-06-internal-guidance-audit.md

@@ -0,0 +1,35 @@
+# Onboarding ONB-06: internal guidance audit
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-system.onb-06-internal-guidance-audit`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Method
+
+Audit every applicable B2C App Builder and internal B2C guidance source against this product's onboarding decisions, recording each rule, its evidence class, and a pass/partial/fail/outdated/conflict/not-applicable verdict, and resolve any conflict or outdated rule explicitly rather than leaving it ambiguous. Write the audit to product/onboarding/graph/ONB-06-internal-guidance-audit.md; check:onboarding-evidence-onb-06 rejects a packet with no real prose findings.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product/onboarding/graph/ONB-02-evidence-plan.md`
+
+Consult when relevant: None declared.
+
+Managed outputs: `product/onboarding/graph/ONB-06-internal-guidance-audit.md`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-evidence-onb-06`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |

package/agents/skills/b2c-design-onboarding/references/workflow.experience.onboarding-system.onb-07-provider-policy-landscape.md

@@ -0,0 +1,37 @@
+# Onboarding ONB-07: provider and policy landscape
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-system.onb-07-provider-policy-landscape`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Method
+
+Refresh the current RevenueCat/billing surface (SDKs, offerings, entitlements, paywalls, experiments, pending purchases, restore) and PostHog, App Store Connect, and Google Play capability and policy facts by region, recording a technically-possible/policy-permitted distinction, enrollment/disclosure/fee/reporting detail, and a revalidation date for each. Write the findings to product/onboarding/graph/ONB-07-provider-policy-landscape.md; check:onboarding-evidence-onb-07 rejects a stub packet, and stale provider facts here propagate directly into ONB-12's state contract and ONB-17's paywall contract.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product/onboarding/graph/ONB-02-evidence-plan.md`
+
+Consult when relevant: None declared.
+
+Managed outputs: `product/onboarding/graph/ONB-07-provider-policy-landscape.md`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-evidence-onb-07`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [PostHog Agent Tooling](../../../../knowledge/data/posthog-agent-tooling.md) | when the selected PostHog operation needs official agent setup, identity, instrumentation, or verification procedures | `reference.data.posthog-agent-tooling` |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |
+| [RevenueCat Agent Tooling](../../../../knowledge/money/revenuecat-agent-tooling.md) | when the selected RevenueCat operation needs official agent setup, identity, instrumentation, or verification procedures | `reference.money.revenuecat-agent-tooling` |

package/agents/skills/b2c-design-onboarding/references/workflow.experience.onboarding-system.onb-08-motion-research.md

@@ -0,0 +1,35 @@
+# Onboarding ONB-08: motion research
+
+<!-- Generated task skill. Edit catalog owners, then run render:task-skills. -->
+
+Canonical workflow: `workflow.experience.onboarding-system.onb-08-motion-research`. This is a reference contract, not an executable plan. Managed eligibility, ordering, authority, and acceptance remain with the installed business runtime.
+
+## Method
+
+Research interaction and motion for the surfaces actually selected. Resolve applicability from studio/seed/business.json `job` (conversion purpose) and `interaction` (technique) on each listed surface, implemented scroll-linked hooks attributed to that surface's file stem under growth/landing, growth/funnel, or web when that landing is active, and growth/landing/surface-contract.json `scrollytelling.applicable` when that file exists; do not infer a technique or conversion job from purpose prose or from packet sentences. Not selected, selected but unavailable, and not yet inspected stay different. A producer sentence that motion research is not applicable cannot override a selected or implemented rich-motion surface. static-document surfaces, and conversion surfaces that did not select or implement scroll-linked interaction, need semantic content, legibility, accessibility, and truthful claims — not invented scroll choreography or 60fps shot IDs. A conversion job on a scroll-linked surface keeps both conversion measurement and motion proof. standard-transition native UI needs usable feedback, interruption, accessibility, and applicable performance, not a paid catalog. When interaction is bespoke-motion or scroll-linked, or that page already implements scroll-linked hooks, and strategy/TOOL_DECISIONS.md records the 60fps MCP as ready, active, or connected without a fallback route, use search_shots, get_shot, get_motion_breakdown, and get_related_shots (motion code only when it clarifies) and record Motion Research rows (target/problem, reference shot ID, adopted principle, implementation/haptic/interruption behavior, reduced-motion behavior). When that technique is selected and the 60fps MCP is blocked or a fallback, hold with the current selection and authority — do not spend, and do not claim a distilled recipe is equivalent. When a listed surface omits `interaction`, hold. An empty studio inventory means 60fps is not selected. Translate adopted principles into this product's own behavior and target framework, not a copy of the reference. Write the packet to product/onboarding/graph/ONB-08-motion-research.md; check:onboarding-evidence-onb-08 rejects a stub packet and an applicability hold.
+
+## Inputs and outputs
+
+For advisory work, use the evidence supplied and return findings. Do not create runtime state, prerequisite artifacts, or a whole business to satisfy these managed-workspace paths. Missing evidence remains explicit.
+
+Managed inputs: `product/onboarding/graph/ONB-02-evidence-plan.md`, `product.yaml`, `strategy/TOOL_DECISIONS.md`
+
+Consult when relevant: `studio/seed/business.json`, `DESIGN.md`
+
+Managed outputs: `product/onboarding/graph/ONB-08-motion-research.md`
+
+## Verification and authority
+
+Declared checks: `check:onboarding-evidence-onb-08`
+
+Additional founder decisions: None specific to this workflow; host permission boundaries still apply.
+
+Run applicable checks with the installed builder's documented target arguments. Report unavailable or unrun checks honestly. A review report, generated file, or successful command does not by itself establish accepted runtime or provider evidence. Do not mark reducer state by hand.
+
+## Select supporting knowledge
+
+Load only the sections needed for the current task. Broad load-when wording does not override accepted scope or an explicit provider binding. Provider procedures are current only at the corresponding selected-provider action. Stop and report an unresolved required source rather than treating its title as delivered guidance.
+
+| Reference | Load when | Stable identity |
+| --- | --- | --- |
+| [Onboarding Conversion](../../../../knowledge/experience/onboarding-conversion.md) | before onboarding quizzes, welcome/splash screens, personalization, attribution questions, demo videos, App Review popups, paywall timing, closing offers, trials, or first-session activation | `reference.experience.onboarding-conversion` |