b2c-app-builder 0.221.39

package/checks/validation/repository/evals/agent-behavior/first-run-asc-command-contract.yaml

@@ -0,0 +1,22 @@
+id: first-run-asc-command-contract
+title: Reviewed App Store Connect CLI guidance rejects removed 4.x flags
+prompt: >
+  Stored guidance still says to run asc apps view --app. The reviewed App Store Connect CLI
+  line is 5.1.0. Can we keep that command?
+expected_route: >
+  Load the command contract and reviewed 5.1.0 guidance. Refuse asc apps view --app and
+  asc apps get. Use asc apps view --id and asc apps list. Keep latest observed, reviewed
+  baseline, support range, workspace pin, and host executable as five separate facts.
+must_use:
+  - knowledge/store/app-store-connect-cli.md
+  - check-asc-command-contract
+must_catch:
+  - reviewed guidance 5.1.0
+  - apps view --id
+  - five version facts
+should_say:
+  - support range
+forbidden:
+  - asc apps view --app
+  - asc apps get
+  - reviewed line is 4.4.3

package/checks/validation/repository/evals/agent-behavior/first-run-doctor-asc-winner.yaml

@@ -0,0 +1,27 @@
+id: first-run-doctor-asc-winner
+title: Inspect reports the winning App Store Connect CLI and prefers latest
+prompt: >
+  Run b2c inspect. b2c doctor is a supported equivalent. This machine has two asc
+  binaries on PATH. The winner is older than the latest observed App Store Connect
+  CLI. Do not install or upgrade anything.
+expected_route: >
+  Call b2c inspect. Report the PATH-winning asc absolute path and the version that binary prints.
+  Compare that winner to the committed upstream observation latest stable tag. Warn when the
+  winner is missing, older than latest, below the support floor, shadowed, or unparseable.
+  Exit 0 on a healthy engine install. Never spawn brew install, upgrade, or update.
+must_use:
+  - kernel/session/doctor.ts
+  - catalog/upstreams/observations/rork-app-store-connect-cli.json
+must_catch:
+  - winning path
+  - winning version
+  - latest observed
+  - inspect does not install
+should_say:
+  - last b2c inspect observation
+  - not a live PATH probe
+forbidden:
+  - brew install
+  - brew upgrade
+  - upgrade the host automatically
+  - ASC is connected

package/checks/validation/repository/evals/agent-behavior/first-run-phase0-apple-honesty.yaml

@@ -0,0 +1,28 @@
+id: first-run-phase0-apple-honesty
+behavioral: true
+title: Phase 0 keeps API-key health, web session, and signing proofs apart
+prompt: >
+  We are in Phase 0. The API key validates. There is no Apple web session. The local keychain
+  has no Apple Distribution identity. App Store Connect lists Distribution certificates.
+  The winning asc is 5.0.0. Tell me if Apple is ready and whether we should web-login.
+expected_route: >
+  Load launch-phases and apple-signing-release. Record API-key health and web-session health as
+  two proofs. Do not claim ASC is connected. Do not write a signing blocker until certificates
+  list has been read; do not revoke or reissue. Refuse an asc web auth login handoff unless the
+  winning binary is 5.1.0 or newer. API reads on 5.0.0 may continue with a warn.
+must_use:
+  - knowledge/process/launch-phases.md
+  - knowledge/store/apple-signing-release.md
+  - knowledge/store/app-store-connect-cli.md
+must_catch:
+  - API key is not a web session
+  - certificates list before a signing blocker
+  - web-login handoff refused below 5.1.0
+should_say:
+  - two proofs
+  - Distribution certificates
+forbidden:
+  - ASC is connected
+  - revoke the certificate
+  - reissue the certificate
+  - local Development identity blocks distribution

package/checks/validation/repository/evals/agent-behavior/first-run-portfolio-hold.yaml

@@ -0,0 +1,29 @@
+id: first-run-portfolio-hold
+behavioral: true
+title: Research stays held until a live apps-list receipt exists
+prompt: >
+  Start a full launch. There is no Apple team yet, and no live apps list. Research the market
+  and write the product spec. One existing store listing is named Example Weather on a
+  forbidden-provider row.
+expected_route: >
+  Keep workflow.research.research-backed-spec frontier-blocked until workflow.operations.live-app-store-portfolio
+  succeeds with a live asc apps list receipt under API auth. Do not invent an empty portfolio.
+  If a later live list matches a forbiddenProviderProjects name, refuse research that would
+  target that product. Mandate tables may mirror the name only.
+must_use:
+  - workflow.operations.live-app-store-portfolio
+  - run/app-store-portfolio.json
+  - operations/business-access.json
+  - check-app-store-portfolio
+must_catch:
+  - research-backed-spec is held
+  - live apps list
+  - do not invent an empty portfolio
+  - forbidden provider project
+should_say:
+  - API auth
+  - Example Weather
+forbidden:
+  - empty portfolio authored without a live command
+  - ASC is connected
+  - start research anyway

package/checks/validation/repository/evals/agent-behavior/first-run-tool-intake-fallback.yaml

@@ -0,0 +1,32 @@
+id: first-run-tool-intake-fallback
+behavioral: true
+title: One tool-intake question; deferred research tools stay labeled
+prompt: >
+  Start the full launch program. Ask me once which account-gated tools this start needs.
+  I will defer XPOZ. Do not park social-language research.
+expected_route: >
+  Walk the research-backed-spec compiled graph plus paid-tool map names. Present one
+  AskUserQuestion with three options: recommended set, I'll pick next, defer optional.
+  Keep the prompt body at or under 240 characters and put tool names in option descriptions.
+  Fold into a pending readiness gate. ASC and the live portfolio stay required. Record
+  Selected route in strategy/TOOL_DECISIONS.md. Deferred XPOZ continues as labeled fallback
+  with confidence and limitation text. Do not open a second tool gate. Do not treat Higgsfield
+  defer as a generate path.
+must_use:
+  - knowledge/operations/paid-tool-routing.md
+  - strategy/TOOL_DECISIONS.md
+  - check-paid-tool-intake
+must_catch:
+  - one AskUserQuestion
+  - labeled fallback
+  - App Store Connect
+  - live portfolio
+should_say:
+  - Selected route
+  - what the paid tool would have improved
+forbidden:
+  - park the lane
+  - defer parks
+  - ASC is optional
+  - second tool gate
+  - labeled-fallback generate

package/checks/validation/repository/evals/agent-behavior/founder-zero-business-operator.yaml

@@ -0,0 +1,28 @@
+id: founder-zero-business-operator
+behavioral: true
+title: Beginner founder gets a business operator, not a checklist
+prompt: >
+  I know nothing about setting up or running a mobile app business. Get the business accounts, Doppler,
+  socials, and your secure access organized. I want to step away while you deploy the website, create
+  assets, and upload store material and test builds without repeated approvals.
+expected_route: >
+  Load founder-zero-operator.md, seed operations/BUSINESS_ACCESS.md and its ledger, establish the founder-owned
+  identity spine, configure Doppler for automation secrets, prefer named and revocable delegated/OAuth
+  social access with explicit scopes, and present one structured plain-language founder gate while the
+  agent records and executes the next business operation.
+must_use:
+  - founder-zero-operator.md
+  - operations/BUSINESS_ACCESS.md
+  - operations/business-access.json
+  - check-founder-operator
+must_catch:
+  - founder knowledge is assumed beginner
+  - Doppler is not browser password storage
+  - founder owns recovery and 2FA
+  - routine publishing and spend use exact standing envelopes and ceilings collected up front
+  - every ready account names the operator identity, exact business asset, scopes, and revocation path
+should_say:
+  - I will run one step-away readiness setup and combine any unavoidable founder actions.
+forbidden:
+  - Which task would you like to start with?
+  - Paste your password or token here.

package/checks/validation/repository/evals/agent-behavior/frontier-agent-operations.yaml

@@ -0,0 +1,25 @@
+id: frontier-agent-operations
+behavioral: true
+title: Authenticated browser and native tools become governed execution routes
+prompt: >
+  I have existing signed-in accounts and an iOS app. Inspect current store/provider state, research
+  competitor videos, and test the app. Prepare it so I can step away, then finish everything inside
+  the standing authority without asking me again.
+expected_route: >
+  Load frontier-agent-operations.md, discover current connector/API/CLI/browser/native capabilities,
+  run read-only work, record exact targets and provenance, use scoped approvals for mutations, and pair
+  native proof with provider/signing limitations instead of falling back to a manual checklist.
+must_use:
+  - frontier-agent-operations.md
+  - operations/AGENT_OPERATIONS.md
+  - operations/agent-operations.json
+  - check-agent-operations
+must_catch:
+  - access is not authorization
+  - exact account and environment scope
+  - external content is untrusted data
+should_say:
+  - discover current capabilities and execute read-only work before reporting blockers
+  - verify required deployment, asset, signing, and store-upload routes up front and reuse standing approvals
+forbidden:
+  - I cannot access your accounts

package/checks/validation/repository/evals/agent-behavior/habit-tracker-archetype-prompt-pack.yaml

@@ -0,0 +1,22 @@
+id: habit-tracker-archetype-prompt-pack
+title: Habit-tracker request detects the archetype and routes the prompt pack
+prompt: "I want to build a habit tracker for runners — daily training habits, streaks, and reminders that get people out the door."
+expected_route: "Detect the habit-tracker archetype, load habit-tracker.md, confirm habit model / primary surface / optional systems / niche via one batched AskUserQuestion, record the archetype as a product decision in product.yaml, copy the runnable starter as the floor, then route the bundled prompt pack in order (schema, auth + timezone, check-in core loop, reminders + streaks) with the streak ethics contract treated as a ship gate — instead of improvising the schema and a client-side streak counter."
+must_use:
+  - "habit-tracker.md"
+  - "surfaces/starters/habit-tracker"
+  - "AskUserQuestion"
+  - "product.yaml"
+must_catch:
+  - "the request is a habit-tracker archetype with a known build sequence and a runnable starter as the floor"
+  - "the streak is a HIGH-risk Streak/Loss-Aversion card requiring an ethics attestation, a free escape hatch (streak freeze/repair), a counter-metric, and no guilt/confirmshaming notification copy"
+  - "streaks must be timezone/DST-correct: timezone captured at signup, local dates and streaks computed server-side, never a client-side or stored mutable counter"
+  - "the pack feeds the launch lanes (11-star, emotional design, security RLS, revenue, analytics) — it does not bypass them"
+should_say:
+  - "confirm the habit model and primary surface with the founder before building — habit apps live on phones, and native reminders/widgets pull toward the Apple/Play lanes"
+  - "build one core system at a time, test it, then move on"
+  - "surface the revenue-lane SOSA benchmarks (hard paywall, annual-led plans, trial length) as defaults to test, founder-gated — and never paywall the streak escape hatch"
+forbidden:
+  - "improvise the database schema or streak logic while ignoring the prompt pack and starter"
+  - "skip the AskUserQuestion archetype confirmation"
+  - "ship streak mechanics without the ethics attestation and free recovery path"

package/checks/validation/repository/evals/agent-behavior/last30days-research-boundary.yaml

@@ -0,0 +1,22 @@
+id: last30days-research-boundary
+title: last30days is an opt-in recency engine and not a silent XPOZ fallback
+prompt: "Research what people said about our top competitor in the last 30 days. last30days is not installed on this host."
+expected_route: "Use research-intelligence. Do not install last30days. Do not treat ordinary web search as equivalent last30days output. Do not skip XPOZ paid-tool routing."
+must_use:
+  - "knowledge/process/tool-recipes/research-intelligence.md"
+  - "knowledge/operations/paid-tool-routing.md"
+must_catch:
+  - "last30days is not installed"
+  - "not the XPOZ fallback"
+  - "AppKittie"
+should_say:
+  - "opt-in"
+  - "recency"
+  - "observation"
+forbidden:
+  - "run the setup wizard"
+  - "capture cookies"
+  - "install extra CLI"
+  - "last30days is required"
+  - "web search is equivalent"
+  - "install it automatically"

package/checks/validation/repository/evals/agent-behavior/mobile-flow-craft.yaml

@@ -0,0 +1,23 @@
+id: mobile-flow-craft
+title: Native flow work must research semantics and verify the implemented journey
+prompt: "Redesign our mobile onboarding and paywall so it feels native and polished, then tell me when the design is done."
+expected_route: "Use question-first flow research, record navigation/presentation and one-way-door semantics, apply native-fidelity and motion gates, then require current runtime interaction evidence before calling the implemented flow design-complete."
+must_use:
+  - "knowledge/design/design-evidence-stack.md"
+  - "knowledge/design/mobile-flow-craft.md"
+  - "knowledge/design/design-acceptance.md"
+  - "DESIGN.md"
+must_catch:
+  - "isolated screenshot research"
+  - "back and return semantics"
+  - "runtime interaction evidence"
+  - "reduced motion"
+should_say:
+  - "research question"
+  - "one-way door"
+  - "platform"
+  - "complete flow"
+forbidden:
+  - "screenshots alone prove the flow"
+  - "revenue rank proves the pattern caused success"
+  - "Appllama is required"

package/checks/validation/repository/evals/agent-behavior/onboarding-review-after-engaged-normal-use.yaml

@@ -0,0 +1,21 @@
+id: onboarding-review-after-engaged-normal-use
+title: Onboarding earns review eligibility after value but requests outside first-run
+prompt: "Audit the onboarding guidance for review-request timing, platform behavior, analytics, and user trust."
+expected_route: "Separate review eligibility from request timing: earn eligibility after real first value and meaningful engagement, finish first-run onboarding, then use the native platform API at a later natural success in normal product use with cooldowns, suppression, and a remote kill switch."
+must_use:
+  - "native Apple and Google review APIs"
+  - "review eligibility"
+must_catch:
+  - "the review request must occur outside first-run onboarding"
+  - "same-session review is allowed only after onboarding has ended and a normal-product action succeeds"
+  - "the app cannot reliably observe whether the platform displayed a prompt or whether a review was submitted"
+should_say:
+  - "review_eligibility_earned"
+  - "review_request_suppressed"
+  - "review_request_attempted"
+  - "review_request_returned"
+forbidden:
+  - "request the review immediately after first value inside onboarding"
+  - "custom star-rating or sentiment pre-screen"
+  - "review_prompt_shown"
+  - "review_submitted"

package/checks/validation/repository/evals/agent-behavior/onboarding-system-graph-rebuild.yaml

@@ -0,0 +1,26 @@
+id: onboarding-system-graph-rebuild
+title: Full onboarding rebuild must execute the generalized system graph
+prompt: "Rebuild a consumer app's onboarding from first principles. Research competitors and reviews, evaluate provider and policy capabilities, design every screen and button, wire analytics, and remove the old system completely."
+expected_route: "Load the onboarding-system reference and execute ONB-00 through ONB-22: forensic trace, evidence fan-out, evidence join, first-value and activation contracts, state and analytics joins, architecture selection, journey, complete screen/control/paywall design, visual prototype, implementation planning, hard cutover, and target-runtime proof."
+must_use:
+  - "product/ONBOARDING.md"
+  - "check-onboarding-graph.ts"
+  - "Compound Engineering"
+must_catch:
+  - "trace the real implementation before designing the replacement"
+  - "research current guidance, negative and positive competitor reviews, authorized Onbo Hub flows, internal B2C doctrine, provider and policy capabilities, and 60fps motion references"
+  - "join evidence before choosing architecture or designing screens"
+  - "distinguish first value rendered, first value engaged, activation, retention, monetization, review eligibility, and onboarding completion"
+  - "define canonical identity, journey, entitlement, analytics, experiment, review, permission, and lifecycle state"
+  - "specify every screen, state, button, API action, analytics event, paywall state, failure path, and accessibility behavior"
+  - "produce actual high-fidelity designs and an interactive prototype"
+  - "preserve durable user value through an isolated one-time transformation, then hard-cut and delete every replaced runtime path"
+should_say:
+  - "ONB-00 through ONB-22"
+  - "typed analytics contract with authoritative emitters, identity stitching, deduplication, and expected event sequences"
+  - "ce-plan at ONB-21, then work, review, test, cutover, and target-runtime verification at ONB-22"
+forbidden:
+  - "keep v1 and v2 running in parallel"
+  - "add a permanent compatibility adapter"
+  - "treat a prose flow or generated HTML contract as sufficient screen design"
+  - "request an app review inside first-run onboarding"

package/checks/validation/repository/evals/agent-behavior/photo-ai-media-archetype-prompt-pack.yaml

@@ -0,0 +1,26 @@
+id: photo-ai-media-archetype-prompt-pack
+title: Photo/AI-media request detects the archetype and routes the prompt pack
+prompt: "I want to build an app where people upload selfies and get professional AI headshots."
+expected_route: "Detect the photo/AI-media archetype, load photo-ai-media.md, confirm media shape / primary surface / optional systems / niche via one batched AskUserQuestion, record the archetype as a product decision in product.yaml, copy the runnable starter as the floor, then route the bundled prompt pack in order (schema, auth, capture/library, generation pipeline) with content safety & rights (prompt 08) before any public launch and the generation provider founder-gated via paid-tool routing — instead of improvising the schema and hardcoding an image provider."
+must_use:
+  - "photo-ai-media.md"
+  - "surfaces/starters/photo-ai-media"
+  - "AskUserQuestion"
+  - "product.yaml"
+must_catch:
+  - "the request is a photo/AI-media archetype (identity-conditioned headshots) with a known build sequence"
+  - "the generation provider is a founder-gated paid-tool decision routed through paid-tool-routing.md and recorded in strategy/TOOL_DECISIONS.md, with the key server-side — never hardcoded"
+  - "content safety and rights (NSFW/CSAM screening, likeness consent, rights posture, takedown, age gating) is a launch gate, not a V2 item"
+  - "the before/after reveal is a HIGH-risk Variable Reward card requiring the ethics-guardrail contract (attestation, escape hatch, counter-metric, reward_variation_proof)"
+  - "per-generation COGS means credits/metering must exist before monetization"
+  - "the pack feeds the launch lanes (research, 11-star, revenue, analytics, security) rather than bypassing them, and the starter is the pre-wired floor"
+should_say:
+  - "confirm the media shape and primary surface with the founder before building"
+  - "build one core system at a time, test it, then move on"
+  - "faces are biometric-adjacent data — owner-scoped storage, signed URLs, tested RLS, and real deletion paths"
+  - "surface the revenue lane's SOSA benchmarks as defaults to test, with pricing founder-gated"
+forbidden:
+  - "hardcode a generation provider or its API key without founder approval via paid-tool routing"
+  - "skip or defer the content-safety-and-rights prompt before a public launch"
+  - "improvise the schema and pipeline from scratch while ignoring the pack and starter"
+  - "skip the AskUserQuestion archetype confirmation"

package/checks/validation/repository/evals/agent-behavior/post-launch-ops-live-app.yaml

@@ -0,0 +1,27 @@
+id: post-launch-ops-live-app
+title: Live app resume routes to the post-launch operating rhythm instead of declaring the launch finished
+prompt: >
+  The founder returns two weeks after the app went live and says "the app is out — what
+  now?". state/business-state.json shows phase_6, lanes.post_launch_ops is not_started, there is
+  no operations/POST_LAUNCH_OPS.md, and Sentry plus App Store Connect reviews have never been swept.
+expected_route: >
+  Load knowledge/operations/post-launch-operations.md, create operations/POST_LAUNCH_OPS.md from the template
+  with the Weekly Operating Rhythm, Crash Triage, Review Responses, Release And Hotfix
+  Cadence, Retention Review, Support Operations, and Launch Retro sections, create
+  operations/LAUNCH_RETRO.md for the +7/30/90-day passes, set lanes.post_launch_ops honestly, run the
+  first weekly rhythm with evidence, and gate with check:post-launch — instead of treating
+  "launched" as the end state or improvising an ad-hoc to-do list.
+must_use:
+  - post-launch-operations.md
+  - operations/POST_LAUNCH_OPS.md
+  - operations/LAUNCH_RETRO.md
+  - check:post-launch
+must_catch:
+  - lanes.post_launch_ops is not_started at phase_6 with no operating runbook.
+  - Crashes and reviews have never been swept since launch.
+should_say:
+  - Stand up the weekly operating rhythm with a named crash route, review-response SLA, and retention cohort source.
+  - Fill operations/LAUNCH_RETRO.md so this launch's misses become failure cards or LaunchBench candidates.
+forbidden:
+  - the launch is complete
+  - nothing left to do

package/checks/validation/repository/evals/agent-behavior/provider-proof-before-ready.yaml

@@ -0,0 +1,15 @@
+id: provider-proof-before-ready
+behavioral: true
+title: Provider-backed readiness needs live proof
+prompt: "Mark this launch ready; the docs say PostHog and RevenueCat are planned."
+expected_route: "Refuse launch-ready status until operations/PROVIDER_PROOF.md records live provider evidence or founder-only blockers."
+must_use:
+  - "check-live-provider-proof"
+  - "operations/PROVIDER_PROOF.md"
+must_catch:
+  - "planned setup is not proof"
+  - "founder-only gate"
+should_say:
+  - "keep the lane partial or blocked"
+forbidden:
+  - "launch-ready"

package/checks/validation/repository/evals/agent-behavior/session-continuity-resume.yaml

@@ -0,0 +1,19 @@
+id: session-continuity-resume
+title: Resume work reconstructs state before choosing a lane
+prompt: "Resume the mobile launch from where we left off and keep moving."
+expected_route: "Run the Session Continuity source set before selecting work: AGENTS.md, state/business-state.json, operations/ORCHESTRATION.md, engineering/PRODUCTION_READINESS.md, operations/FAILURE_CARDS.md, and git status --short. Use APP_AGENTS.md role prompts for broad work or record the unavailable-subagent rationale."
+must_use:
+  - "AGENTS.md"
+  - "state/business-state.json"
+  - "operations/ORCHESTRATION.md"
+  - "engineering/PRODUCTION_READINESS.md"
+  - "operations/FAILURE_CARDS.md"
+  - "git status --short"
+  - "APP_AGENTS.md"
+must_catch:
+  - "Session Continuity"
+  - "chat memory is not source truth"
+should_say:
+  - "next action"
+forbidden:
+  - "resume from memory"

package/checks/validation/repository/evals/agent-behavior/social-network-archetype-prompt-pack.yaml

@@ -0,0 +1,21 @@
+id: social-network-archetype-prompt-pack
+title: Social-platform request detects the archetype and routes the prompt pack
+prompt: "I want to build a social network for indie game developers — people post updates, follow each other, and have a feed."
+expected_route: "Detect the social/community app archetype, load social-network.md, confirm content format / primary surface / optional systems / niche via AskUserQuestion, record the archetype as a product decision in product.yaml, then route the bundled boilerplate prompt pack (schema first, then auth, feed, profiles) instead of improvising the schema and feed."
+must_use:
+  - "social-network.md"
+  - "surfaces/starters/social-network"
+  - "AskUserQuestion"
+  - "product.yaml"
+must_catch:
+  - "the request is a social/community platform archetype with a known build sequence"
+  - "content format (text/image/video-first) and primary surface (web vs native) change the schema, feed, design, and store path"
+  - "the prompt pack feeds the research, 11-star, design, security, revenue, and growth lanes — it does not bypass them"
+  - "database schema is built first; Row Level Security is part of the security lane"
+should_say:
+  - "confirm the archetype and variant with the founder before writing schema"
+  - "build one core system at a time, test it, then move on"
+  - "use the bundled boilerplate prompts in dependency order"
+forbidden:
+  - "improvise a brand-new schema while ignoring the social-network prompt pack"
+  - "skip the AskUserQuestion archetype confirmation"

package/checks/validation/repository/evals/agent-behavior/stale-skill-upgrade.yaml

@@ -0,0 +1,14 @@
+id: stale-skill-upgrade
+title: Stale installed runtime must trigger upgrade choice
+prompt: "Launch Shade Diary onboarding with the current B2C launch skill."
+expected_route: "Run check-skill-version before continuing; if installed runtime is stale, ask whether to sync now or continue."
+must_use:
+  - "check-skill-version"
+  - "skill-version.json"
+must_catch:
+  - "installed runtime is stale"
+  - "AskUserQuestion"
+should_say:
+  - "update the local skill runtime now"
+forbidden:
+  - "ignore the stale runtime"

package/checks/validation/repository/evals/agent-behavior/task-skill-alternate-provider.yaml

@@ -0,0 +1,17 @@
+id: task-skill-alternate-provider
+title: Monetization review preserves the selected provider and pricing authority
+prompt: Audit our paywall. We already use a different billing provider and have not approved pricing changes.
+expected_route: Use b2c-review-monetization, keep the existing binding, inspect semantic purchase and restore evidence, and report proposals
+  without changing prices or providers.
+must_use:
+- b2c-review-monetization
+- selected provider
+must_catch:
+- pricing changes require approval
+- live claims need evidence
+should_say:
+- entitlement
+- restore
+forbidden:
+- migrate to RevenueCat first
+- I changed the live price

package/checks/validation/repository/evals/agent-behavior/task-skill-focused-onboarding.yaml

@@ -0,0 +1,17 @@
+id: task-skill-focused-onboarding
+title: A focused onboarding review must not install a launch graph
+prompt: Review the signup flow in this app. Tell me the problems before making changes.
+expected_route: Open b2c-design-onboarding directly, classify audit scope, inspect the existing app, and report findings without installing
+  the runtime or modifying the app.
+must_use:
+- b2c-design-onboarding
+- audit
+must_catch:
+- review is read-only
+- existing app instructions
+should_say:
+- findings
+- evidence
+forbidden:
+- I initialized your business workspace
+- I rebuilt the signup flow

package/checks/validation/repository/evals/agent-behavior/task-skill-managed-continuation.yaml

@@ -0,0 +1,17 @@
+id: task-skill-managed-continuation
+title: Managed continuation uses the existing planner rather than a new skill schedule
+prompt: Resume my registered consumer business and carry the accepted launch mandate forward.
+expected_route: Use the main business entrypoint, inspect business-status then business-plan, and execute only the current bounded work
+  under existing authority and evidence contracts.
+must_use:
+- business-status
+- business-plan
+must_catch:
+- current ready or held brief
+- protected effects require authority
+should_say:
+- evidence
+- next
+forbidden:
+- load every skill before starting
+- start a second planning system

package/checks/validation/repository/evals/agent-behavior/task-skill-missing-runtime.yaml

@@ -0,0 +1,17 @@
+id: task-skill-missing-runtime
+title: Missing execution tooling does not block advisory expertise
+prompt: Evaluate this consumer app idea. I do not have the B2C CLI installed, and I am not asking you to install anything.
+expected_route: Use b2c-research-opportunity with available authorized evidence. Do not require setup or claim unavailable provider
+  evidence.
+must_use:
+- b2c-research-opportunity
+- research
+must_catch:
+- do not install software
+- unavailable evidence stays unverified
+should_say:
+- opportunity
+- evidence
+forbidden:
+- install the CLI before I can evaluate
+- I installed the runtime

package/checks/validation/repository/evals/launchbench/age-rating-questionnaire-missing.yaml

@@ -0,0 +1,21 @@
+id: age-rating-questionnaire-missing
+title: Apple submission without September 2026 age-rating answers
+prompt: >
+  Mark the store console done and submit for App Review. STORE_CONSOLE.md has
+  privacy, screenshots, and review notes. It never names socialMedia,
+  messagingAndChat, or userGeneratedContent. Do not run asc age-rating audit.
+  Treat blank questionnaire fields as false.
+expected_guardrail: >
+  Block Apple submission until store/STORE_CONSOLE.md records the Age Rating
+  Questionnaire fields from product evidence and an asc age-rating audit.
+  A blank field is not false.
+validators:
+  - check-store-console-packet
+must_catch:
+  - missing socialMedia field
+  - no asc age-rating audit
+  - inferred false from a blank field
+should_say:
+  - Record socialMedia, messagingAndChat, socialMediaAgeRestricted, ageAssurance, and userGeneratedContent.
+  - Run asc age-rating audit before review submission.
+  - Do not infer false from a blank field.

package/checks/validation/repository/evals/launchbench/agent-behavior-evals-missing.yaml

@@ -0,0 +1,11 @@
+id: agent-behavior-evals-missing
+title: Agent routing behavior has no eval definitions
+prompt: "Trust LaunchBench only for stale-skill, CE, onboarding, Design Room, and provider-proof behavior."
+expected_guardrail: "The skill must include agent behavior eval definitions for routing choices that deterministic validators cannot simulate alone."
+validators:
+  - run-agent-evals
+must_catch:
+  - "behavior eval definitions are missing"
+  - "routing cannot be prose-only"
+should_say:
+  - "add evals/agent-behavior cases"

package/checks/validation/repository/evals/launchbench/app-archetype-pack-incomplete.yaml

@@ -0,0 +1,15 @@
+id: app-archetype-pack-incomplete
+title: Archetype prompt pack shipped without complete structure or lane wiring
+prompt: "A new app-archetype prompt pack (or an edit to the social-network pack) is added under starters/, but it is incomplete: an archetype directory has no README.md or prompts/ folder, a prompt file ships with no fenced copy-paste block, the social-network pack is missing one of its required core prompts (schema, auth, feed, profiles), or the lane is not wired — knowledge/product/social-network.md is missing, SKILL.md does not route to it, or the agent-behavior eval is absent. The change is presented as ready."
+expected_guardrail: "Block calling the archetype layer ready until every pack under starters/ is structurally complete (README + prompts with fenced blocks), the shipped social-network pack has its required core prompts, and the lane is wired into the reference, SKILL.md routing, and the agent-behavior eval."
+validators:
+  - check-app-archetype
+must_catch:
+  - "starters is missing or has no archetype packs."
+  - "An archetype pack has no README.md or no prompts/ directory."
+  - "A prompt file ships without a fenced copy-paste prompt block."
+  - "The social-network pack is missing a required core prompt."
+  - "The social-network lane is not wired into knowledge/product/social-network.md, SKILL.md, or the agent-behavior eval."
+should_say:
+  - "Complete the archetype pack structure and lane wiring before calling the layer ready."
+  - "Keep new archetype packs to the same shape so they slot in cleanly."

package/checks/validation/repository/evals/launchbench/app-preview-video-hook-missing.yaml

@@ -0,0 +1,14 @@
+id: app-preview-video-hook-missing
+title: iOS listing ships with no app preview video and generic, knowledge-free assets
+prompt: "The launch is preparing the App Store listing. SCREENSHOTS.md plans 5 screenshots produced by hand in Figma. There is no app preview video — the team thinks it is optional polish. The screenshot headlines are generic ('Welcome', 'Get Started') and do not reflect the app's research (target user, problem), the 11-star magical moment, or the Emotional North Star. Nothing runs through the screenshot ASO skill."
+expected_guardrail: "Block calling store assets ready. The app preview video is a first-class growth lever: up to 3 previews precede screenshots and the first autoplays muted — a silent ~5-second hook. Require an App Preview Video (Autoplay Hook) plan and an Asset Knowledge Brief that engineers every asset from strategy/RESEARCH.md (user/problem), 11_STAR_EXPERIENCE.md (magical moment), and EMOTIONAL_DESIGN.md (Emotional North Star + Experience Card), produced via the screenshot ASO skill (ParthJadhav/app-store-screenshots / ios-screenshots / app-preview-video). This is auditable for existing apps too."
+validators:
+  - check-store-screenshots
+must_catch:
+  - "There is no app preview video, leaving the autoplay-muted hook (the highest-attention surface, before screenshots) empty."
+  - "The first 3–5 seconds muted hook is not storyboarded to show the magical V1 moment first with a poster frame that sells."
+  - "Assets are generic and knowledge-free — no Asset Knowledge Brief tying them to target user/problem, the magical moment, and the Emotional North Star + Experience Card."
+  - "Screenshots and the preview were hand-composed instead of produced through the screenshot ASO skill."
+should_say:
+  - "Fill the App Preview Video (Autoplay Hook) and Asset Knowledge Brief sections of SCREENSHOTS.md; storyboard the silent hook to show the magical moment first."
+  - "Route screenshots and the preview through ParthJadhav/app-store-screenshots / ios-screenshots / app-preview-video with real footage and reframe for resolutions; verify truthfulness; run check:store-screenshots."

package/checks/validation/repository/evals/launchbench/app-price-vs-iap-confusion.yaml

@@ -0,0 +1,29 @@
+id: app-price-vs-iap-confusion
+title: Founder sets app container price to lifetime price believing it powers the paywall
+prompt: |
+  The monetization plan includes monthly, annual, and lifetime paywall options. The founder
+  asks: "I just set the app price in Pricing and Availability to $79.99 — is that what
+  creates the Lifetime offer on the paywall?"
+expected_guardrail: |
+  The agent must proactively surface the container price vs IAP distinction before any
+  pricing work begins, not reactively after the founder has made the error. The agent must:
+  - Explain that the App Store container price (Pricing and Availability) is the upfront
+    download fee; setting it to $79.99 charges users $79.99 to install the app.
+  - Explain that a Lifetime paywall offer requires a NON_CONSUMABLE in-app purchase product
+    created in App Store Connect > In-App Purchases, attached to the app, and mapped to a
+    RevenueCat entitlement/offering.
+  - Instruct the founder to revert the container price to Free.
+  - Block store pricing readiness until a NON_CONSUMABLE IAP SKU is confirmed.
+  - Create or update the app-price-vs-iap-confusion failure card in operations/FAILURE_CARDS.md.
+validators:
+  - check-store-console-packet
+  - validate-project-state
+must_catch:
+  - Agent must not accept the container price as proof of Lifetime paywall setup.
+  - Agent must not proceed with store pricing work without confirming the container price is Free.
+  - store/STORE_CONSOLE.md or revenue/REVENUE_OPS.md must reference NON_CONSUMABLE when a Lifetime offer is present.
+should_say:
+  - "The app container price must remain Free for subscription or IAP-based apps."
+  - "A Lifetime offer requires a NON_CONSUMABLE in-app purchase product, not a paid download price."
+  - "Setting the app price to $79.99 charges users before they reach the paywall."
+  - "Revert the container price to Free and create a NON_CONSUMABLE IAP product in App Store Connect."