b2c-app-builder 0.221.39

package/ACKNOWLEDGMENTS.md

@@ -0,0 +1,168 @@
+# Acknowledgments
+
+B2C App Builder is built on open-source work and adapts published methods. This file credits the projects and people that power it. It lists actual use or adaptation only, never candidates that were merely inspected. Legal notices are separate, in [THIRD_PARTY_NOTICES.md](THIRD_PARTY_NOTICES.md).
+
+## Open-source projects we run or wrap
+
+**App Store Connect CLI (asc)**
+
+- Original author: Rudrank Riyam, original author and copyright holder (LICENSE, 2026)
+- Current maintainer: Rork (github.com/rorkai), maintaining organization and release publisher
+- License: MIT (verified)
+- Canonical URL: https://github.com/rorkai/App-Store-Connect-CLI
+- Contributes: Automates App Store Connect reads, validation, dry runs, and founder-gated store actions through the installed asc executable. The builder adapts its command procedures and refuses the commands it does not support.
+- How we consume it:
+  - external-executable: The `asc` executable found on the host PATH runs App Store Connect reads, validation, dry runs, and founder-gated store actions. It is not package-managed by the builder and no workspace pin freezes it.
+  - adapted-method: Command procedures, the auth ladder, safe automation boundaries, and the verified command cookbook are reauthored from the CLI README, local `asc --help` output, and release notes into the builder's own vocabulary.
+
+## Methods and guidance we adapted
+
+**Amir Mushich Brand System Skill**
+
+- Original author: Amir Mushich, original author
+- License: CC-BY-4.0 (verified)
+- Canonical URL: https://github.com/amirmushichge/brand-system-skill
+- Contributes: Provider-neutral brand kit and reference roles. Reauthored for existing design authorities; original author and license retained. No endorsement or demonstrated uplift claimed.
+- How we consume it:
+  - adapted-method: Provider-neutral brand kit and reference roles. No source executable or assets imported.
+
+**Amir Mushich Cinematic Scroll Prompt Kit**
+
+- Original author: Amir Mushich, original author
+- License: MIT (verified)
+- Canonical URL: https://github.com/amirmushichge/cinematic-scroll-prompt-kit
+- Contributes: Conditional layer planning and forward/reverse checkpoints. Reauthored for existing design authorities; original author and license retained. No endorsement or demonstrated uplift claimed.
+- How we consume it:
+  - adapted-method: Conditional layer planning and forward/reverse checkpoints. No source executable or assets imported.
+
+**Appllama agent skills**
+
+- Original author: Antmind Ventures Private Limited, original author or copyright holder identified in the reviewed source
+- Current maintainer: Appllama (github.com/Appllama), current upstream maintainer
+- License: MIT (verified)
+- Canonical URL: https://github.com/Appllama/appllama-skills
+- Contributes: Selected Appllama agent-skill methods inform knowledge/design/mobile-flow-craft.md. The app-design and usage skills remain opt-in companion packs. Original authors and license notices are retained; no live integration is claimed.
+- How we consume it:
+  - adapted-method: Selected cross-platform flow-research, navigation-semantics, native-fidelity, anti-generic, motion, and verification methods are reauthored in knowledge/design/mobile-flow-craft.md. No whole skill, executable, native dependency, or asset was installed by this contribution.
+  - selected-skill-guidance: The Expo/React Native app-design skill and the provider-usage skill are named as opt-in companion packs. Installation happens only with founder approval. No SKILL.md is auto-installed as a top-level builder skill, and Appllama MCP is never a default dependency.
+
+**ASO and App Marketing Skills**
+
+- Original author: Erencan, original author or copyright holder identified in the reviewed source
+- License: MIT (verified)
+- Canonical URL: https://github.com/Eronred/aso-skills
+- Contributes: Selected ASO and App Marketing Skills methods inform knowledge/store/screenshot-toolchain.md, knowledge/store/app-store-listing-prep.md, knowledge/process/tool-recipes/growth-and-store-routing.md. Original authors and license notices are retained; no live integration is claimed.
+- How we consume it:
+  - adapted-method: Selected published methods and tool procedures are adapted into existing builder guidance. No whole skill, executable, native dependency or asset was installed by this contribution.
+
+**serve-sim**
+
+- Original author: Evan Bacon, project author identified by reviewed package metadata; no project-specific copyright line in root LICENSE
+- License: Apache-2.0 (verified)
+- Canonical URL: https://github.com/EvanBacon/serve-sim
+- Contributes: Selected serve-sim methods inform knowledge/engineering/xcodebuildmcp-testing.md, knowledge/process/tool-recipes/device-capture-and-proof.md. Original authors and license notices are retained; no live integration is claimed.
+- How we consume it:
+  - adapted-method: Selected published methods and tool procedures are adapted into existing builder guidance. No whole skill, executable, native dependency or asset was installed by this contribution.
+
+**last30days agent skill**
+
+- Original author: Matt Van Horn, original author or copyright holder identified in the reviewed source
+- Current maintainer: mvanhorn (github.com/mvanhorn), current upstream maintainer
+- License: MIT (verified)
+- Canonical URL: https://github.com/mvanhorn/last30days-skill
+- Contributes: Selected last30days recency-research methods inform knowledge/process/tool-recipes/research-intelligence.md. The host skill remains an opt-in companion pack. Original authors and license notices are retained; no live integration is claimed.
+- How we consume it:
+  - adapted-method: Selected recency-window, engagement ranking, entity resolution, and honest source-coverage methods are reauthored in knowledge/process/tool-recipes/research-intelligence.md. No whole skill, executable, native dependency, or asset was installed by this contribution.
+  - selected-skill-guidance: The last30days skill is named as an opt-in companion pack. Installation happens only with founder approval. No SKILL.md is auto-installed as a top-level builder skill, and the engine is never a default provider.
+
+**App Store Screenshots**
+
+- Original author: Parth Jadhav, original author or copyright holder identified in the reviewed source
+- License: MIT (verified)
+- Canonical URL: https://github.com/ParthJadhav/app-store-screenshots
+- Contributes: Selected App Store Screenshots methods inform knowledge/store/screenshot-toolchain.md, knowledge/store/app-store-listing-prep.md, knowledge/store/store-console-workflow.md. Original authors and license notices are retained; no live integration is claimed.
+- How we consume it:
+  - adapted-method: Selected published methods and tool procedures are adapted into existing builder guidance. No whole skill, executable, native dependency or asset was installed by this contribution.
+
+**PostHog wizard**
+
+- Original author: PostHog, publishing organization and copyright holder per root LICENSE
+- License: MIT (verified)
+- Canonical URL: https://github.com/PostHog/wizard
+- Contributes: Selected procedures inform knowledge/data/posthog-agent-tooling.md. Original notices are retained; no whole-project integration or live support is claimed.
+- How we consume it:
+  - adapted-method: Adapt selected setup and verification procedures as subordinate guidance; no upstream scripts, SDK binaries, or top-level skills are installed.
+
+**RevenueCat AI Toolkit**
+
+- Original author: RevenueCat, publishing organization and copyright holder per root LICENSE
+- License: MIT (verified)
+- Canonical URL: https://github.com/RevenueCat/ai-toolkit
+- Contributes: Selected procedures inform knowledge/money/revenuecat-agent-tooling.md. Original notices are retained; no whole-project integration or live support is claimed.
+- How we consume it:
+  - adapted-method: Adapt selected setup and verification procedures as subordinate guidance; no upstream scripts, SDK binaries, or top-level skills are installed.
+
+**RevenueCat CLI**
+
+- Original author: RevenueCat, publishing organization and copyright holder per root LICENSE
+- License: MIT (verified)
+- Canonical URL: https://github.com/RevenueCat/cli
+- Contributes: Selected procedures inform knowledge/money/revenuecat-agent-tooling.md. A typed process adapter wraps a trusted host rc/revenuecat binary for reviewed operations. Original notices are retained; live host execution remains unproven.
+- How we consume it:
+  - adapted-method: Adapt selected setup and verification procedures as subordinate guidance; no upstream scripts, SDK binaries, or top-level skills are installed.
+  - external-executable: The host `rc` or `revenuecat` binary, when it identifies as RevenueCat CLI 0.1.1, runs selected typed operations through adapters/providers/revenuecat. The builder does not vendor the Go application or fetch an unpinned latest package at runtime.
+
+**App Store Connect CLI skills (asc skill pack)**
+
+- Original author: Rudrank Riyam, original author and copyright holder (LICENSE, 2026)
+- Current maintainer: Rork (github.com/rorkai), maintaining organization; community-maintained, unofficial, not affiliated with Apple
+- License: MIT (verified)
+- Canonical URL: https://github.com/rorkai/app-store-connect-cli-skills
+- Contributes: Community skill pack for the asc CLI. The builder routes selected skills by name as subordinate guidance and installs the pack only with founder approval.
+- How we consume it:
+  - selected-skill-guidance: Selected skill names (asc-cli-usage, asc-metadata-sync, asc-shots-pipeline, asc-screenshot-resize, asc-release-flow, asc-submission-health, asc-revenuecat-catalog-sync, asc-analytics-reports, asc-ad-hoc-distribution, and the rest of the routed list) are loaded as subordinate reference material when a store workflow selects them. Installation happens only with founder approval through `asc install-skills` (pinned by the CLI to a reviewed commit) or `npx skills add`. No SKILL.md is auto-installed as a top-level builder skill.
+
+**SnapshotPreviews**
+
+- Original author: Emerge Tools, original author or copyright holder identified in the reviewed source
+- Current maintainer: Sentry (getsentry), current upstream maintainer
+- License: MIT (verified)
+- Canonical URL: https://github.com/getsentry/SnapshotPreviews
+- Contributes: Selected SnapshotPreviews methods inform knowledge/engineering/xcodebuildmcp-testing.md, knowledge/process/tool-recipes/device-capture-and-proof.md. Original authors and license notices are retained; no live integration is claimed.
+- How we consume it:
+  - adapted-method: Selected published methods and tool procedures are adapted into existing builder guidance. No whole skill, executable, native dependency or asset was installed by this contribution.
+
+**XcodeBuildMCP**
+
+- Original author: Cameron Cooke, original author or copyright holder identified in the reviewed source
+- Current maintainer: Sentry (getsentry), current upstream maintainer
+- License: MIT (verified)
+- Canonical URL: https://github.com/getsentry/XcodeBuildMCP
+- Contributes: Selected XcodeBuildMCP methods inform knowledge/engineering/xcodebuildmcp-testing.md, knowledge/process/tool-recipes/device-capture-and-proof.md. Original authors and license notices are retained; no live integration is claimed.
+- How we consume it:
+  - adapted-method: Selected published methods and tool procedures are adapted into existing builder guidance. No whole skill, executable, native dependency or asset was installed by this contribution.
+
+## Optional managed services
+
+**Layers Growth MCP and CLI**
+
+- Original author: Layers, copyright holder of the public MCP description repository (LICENSE, 2026) and operator of the hosted service
+- Current maintainer: Layers (github.com/layers), maintaining organization
+- License: MIT (verified)
+- Canonical URL: https://github.com/layers/mcp
+- Contributes: Optional managed growth provider. The builder ships a contribution package and adapter with fixtures only; it does not sign a business into Layers and its tests make no live call.
+- How we consume it:
+  - remote-service: The hosted MCP endpoint is reached through the `layers` CLI stdio proxy after a separately authorized business signs in. The builder ships a contribution package (examples/extensions/layers-growth) that declares draft creative generation plus job and result reads, an adapter with a fake transport, and an onboarding contract. No business is onboarded by the builder and no live call is made by its tests.
+  - adapted-method: Tool names, setup steps, credit rules, and the quoteOnly behavior in examples/extensions/layers-growth/integration.md are reauthored from the layers/mcp README (MIT); the retained notice is examples/extensions/layers-growth/LICENSE-layers-mcp.txt
+
+**TypeSafe AI (Jev / System One)**
+
+- Original author: TypeSafe AI, publishing organization for docs, hosted System One API, and client SDKs
+- License: MIT (verified)
+- Canonical URL: https://github.com/typesafe-ai/typesafe-sdk-python
+- Contributes: Candidate semantic-assessment provider (TypeSafe System One / Jev). Brigade records qualification mapping and retains the Python SDK MIT notice; it does not install the SDK, call the API, or claim live support in this unit.
+- How we consume it:
+  - remote-service: Qualify the hosted System One evaluation API (POST /v1/systemone) as a candidate semantic-assessment provider against landed SQ-01 contracts. Paper qualification only — no live call, no Brigade package dependency, no credential invent, no adapter transport.
+  - adapted-method: Reauthor native Choice/Score/Noul (+ batch) shapes from official public docs into a Brigade-owned mapping matrix targeting contracts/semantic/. No upstream SDK types are imported into policy. Independent fixtures preserve official-doc JSON examples with provenance; they are not adapter-authored expected-native captures.
+
+Generated by tooling/render-credits.ts from catalog/upstreams. Edit the manifests, not this file.

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Eduardo Muth Martinez
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,252 @@
+<div align="center">
+
+# Brigade
+
+**Build the business around the app.**
+
+Primitives for creating, launching, and improving consumer-app businesses.
+
+[Get started](#get-started) · [Documentation](docs/README.md) · [Architecture](docs/north-star-architecture.md) · [Contribute](CONTRIBUTING.md)
+
+</div>
+
+![Sheet 1. The kitchen layout, six stations and a pass, with the business going out the door.](docs/assets/business-primitives.svg)
+
+A consumer business needs more than an app that compiles. It needs a product
+people want, a distinctive experience, a working funnel, and reliable
+monetization. It needs a way to learn what works and improve.
+
+Brigade gives AI agents a shared foundation for that whole business.
+It is laid out like a restaurant kitchen. Stations are the responsibilities
+every consumer-app business has, and they do not change with the app.
+Capabilities define what each station must produce. Providers implement them.
+Recipes arrange the work into creation and operating loops. Selected knowledge
+is the mise en place that helps the agent decide well. You are the executive
+chef. You choose the opportunity and the product, set the bar, and let your
+agents cook. The [ethos](docs/ethos.md) explains the stations and who runs them.
+
+Skills and sourced knowledge are part of that foundation, and this repository
+ships a lot of both. Each one describes how to do a single job well. What it
+leaves open is which job comes next, what the last job decided, which provider
+will run this one, whether the result holds up, and what happens when it fails.
+Capabilities, recipes, planning, bounded execution, evidence, and recovery
+answer those questions. They are what keep several specialized agents and
+providers working on the same consumer business rather than on disconnected
+pieces of it.
+
+You run all of this yourself. You bring the agent and model, the infrastructure,
+the provider accounts and credentials, and the environment the work executes in.
+The method is here in full under the MIT licence: the capabilities, recipes,
+knowledge, planning, execution, review, evidence, recovery, and the extension
+contracts for adding your own.
+
+The goal is to move human attention from building one app to designing and
+testing businesses. Produce one excellent consumer business first. Then reuse
+its foundations for a different approach to the same market, with comparable
+evidence.
+
+## Choose how to use it
+
+Use [a focused task skill](agents/skills/README.md) for research, an onboarding review, or a monetization question. Read and apply the expertise without creating a workspace. Use the main business skill when you need coordinated, durable creation or operation through status, plan, execution, and evidence. Both use the same catalog and knowledge.
+
+[Browse the knowledge by business area](knowledge/README.md).
+
+## Get started
+
+Use Node.js 24.
+
+**Today the package is not on the npm registry yet** (`npm view b2c-app-builder`
+returns 404). Use a source checkout until a maintainer publishes the first
+version. The npm install line below is the intended consumer path **after**
+that first publish — not a working registry command today.
+
+```bash
+git clone https://github.com/Clueless-Creations/b2c-app-builder.git
+cd b2c-app-builder
+npm ci
+npm run setup
+```
+
+From a checkout, setup runs `npm link`, so `b2c` resolves to that checkout.
+Setup creates the local workspace registry and prints the MCP registration
+command for Claude Code, Cursor, and Codex. Register that local server as
+`b2c-local`. Hosted knowledge is a separate `b2c-hosted` connection and cannot
+run a workspace.
+
+After the first authorized publish, the consumer path is:
+
+```bash
+npm install -g b2c-app-builder
+b2c setup
+```
+
+The portable MCP form (post-publish) needs no global install:
+
+```bash
+claude mcp add --scope user b2c-local -- npx -y b2c-app-builder
+```
+
+A leftover `b2c-app-builder` client name is not a third capability. Read the
+connection receipt. See leftover names in the
+[package guide](docs/guides/runtime-package.md#mcp). Setup never edits Claude,
+Cursor, or Codex config.
+
+Either way, start with the ordinary business path. The [`b2c-app-builder` skill](SKILL.md)
+routes that work. `b2c --help` groups commands by task; see
+[Find a command](docs/guides/runtime-package.md#find-a-command) rather than a
+second command table here. Catalog discovery and composition preview are optional later.
+
+For a new business, create and register its planning workspace in one command:
+
+```bash
+b2c business-create --workspace my-app --directory /absolute/path/to/my-app --name "My App" --hypothesis "The consumer problem to research" --mandate-file ./brief.md --json
+b2c business-status --workspace my-app --json
+b2c business-plan --workspace my-app --json
+```
+
+The target must be empty or absent. For an existing planning or runtime workspace,
+use `b2c workspaces register` to adopt it, then resume through status and plan.
+Creation records a hypothesis; research and an explicit product decision precede
+initialization. Direct `--mandate` is for short requests. `--mandate-file` preserves
+the complete founder brief in `operations/FOUNDER_BRIEF.md`.
+
+Continue from the canonical brief in `operations/FOUNDER_BRIEF.md`. `operations/LAUNCH_PROGRAM.md`
+is the derived program view and does not replace that source.
+
+The [business-building guide](docs/guides/build-a-business.md) walks through
+the product workflow, design decisions, and release boundaries.
+
+The public lifecycle creates a planning workspace, initializes its accepted
+product with the 99-workflow complete-business recipe, plans eligible work, and
+runs bounded sessions through the existing executor. Evidence reports separate
+current acceptance from missing or stale proof. Interrupted requests close through revision-checked recovery after uncertain effects are reconciled. See the [lifecycle contract](docs/public-interface.md#business-lifecycle)
+for commands, revisions, request replay, and recovery.
+
+To inspect the catalog or preview a composition without creating a workspace:
+
+```bash
+b2c catalog --json
+b2c compose --config contracts/public-api/examples/subscription-app.json --json
+```
+
+The preview resolves each operation to a provider and lists any blockers.
+This declaration preview reports `canApply: false`. To activate an installed
+package recipe, use the separate revision-checked
+[composition plan and activation commands](docs/guides/composition-activation.md).
+
+## What it covers
+
+<!-- catalog-generated:start business-areas -->
+| Area | Station | What the system helps an agent do |
+| --- | --- | --- |
+| [Opportunity](knowledge/README.md#opportunity) | Prep & design, menu planning | Research users, competitors, demand, and a defensible product hypothesis. |
+| [Product](knowledge/README.md#product) | Prep & design | Define the promise, first value, core loop, complete scope, and success measures. |
+| [Experience](knowledge/README.md#experience) | Prep & design | Develop a distinct identity, onboarding, interaction, motion, accessible states, and user-facing words. |
+| [Engineering](knowledge/README.md#engineering) | The hot line | Build and release native and web surfaces with explicit contracts, runtime verification, privacy, and security. |
+| [Revenue and growth](knowledge/README.md#revenue-and-growth) | Front of house | Establish subscriptions, acquisition, funnels, attribution, and lifecycle work. |
+| [Learning and operations](knowledge/README.md#learning-and-operations) | The pass and the office | Inspect evidence, plan improvements, support users, coordinate work, and maintain the business. |
+<!-- catalog-generated:end business-areas -->
+
+Each area has workflow and knowledge coverage. See [Status](#status) for what
+is implemented today.
+
+## How it fits together
+
+| Primitive  | What it is                                                                 | In the kitchen              | What you can change                                       |
+| ---------- | -------------------------------------------------------------------------- | --------------------------- | --------------------------------------------------------- |
+| Capability | A business responsibility with operation semantics and acceptance criteria | The equipment at a station  | Which responsibilities your business needs                |
+| Provider   | An implementation of specific operations, with declared support and limits | The purveyor                | Vendor, tool, native stack, or execution environment      |
+| Recipe     | A configurable creation or operating loop                                  | How you run service         | Sequence, review policy, parameters, and product approach |
+| Knowledge  | Sourced guidance, loaded in bounded amounts                                | Mise en place               | The expertise selected for the task, provider, and recipe |
+| Evidence   | Observations tied to a claim, artifact, and environment                    | The plate check at the pass | The tool that produces the observation, not its meaning   |
+
+A provider swap does not redefine what an entitlement means. A different
+analytics tool does not redefine activation. A new recipe reuses the same
+execution and evidence machinery. Shared foundations should make distinctive
+apps easier to build. Builders keep control of product, brand, business model,
+and operating loop.
+
+A composition binds a recipe to a target. It can override the provider for a
+single operation:
+
+```yaml
+apiVersion: b2c/v1
+recipe:
+  id: b2c/subscription-app
+  version: 1.0.0
+target:
+  platform: ios
+  runtime: swiftui
+bindings:
+  b2c/monetization.present-paywall:
+    provider:
+      id: b2c/revenuecat
+      version: 1.0.0
+    connection: connection:paywalls
+```
+
+Save this as `b2c.yaml` and run `b2c compose --config b2c.yaml --json`. Every
+other operation uses the recipe default. Connection values are names of
+separately managed connections, never credentials. The
+[architectural shape](docs/north-star-architecture.md#architectural-shape)
+shows how a request moves through entrypoints, contracts, composition,
+execution, and evidence.
+
+![Sheet 3. One station in detail, with its operation, bound provider, loaded knowledge, and returned evidence.](docs/assets/station-detail.svg)
+
+Sheet 3 draws that binding as one station. The purveyor can change, and the
+operation keeps its meaning.
+
+[Public interface](docs/public-interface.md) · [Generated schemas](contracts/public-api/REFERENCE.md) · [Extension guide](docs/guides/extend-the-system.md)
+
+## Operate the app
+
+Mobile app operation is a capability of its own. Agents use it to walk
+customer journeys, reproduce bugs, review design, and capture screenshots or
+demo footage. The default provider is the agent host's native simulator
+tooling. MobAI or another provider can be selected when its coverage fits the
+task better. The [mobile app operation guide](docs/guides/mobile-app-operation.md)
+covers provider selection and what counts as evidence.
+
+## Example quality
+
+[Tuck](examples/tuck/README.md) shows real app and web captures, including trip planning,
+packing, and the landing page. The gallery links capture provenance and
+verification limits.
+
+## Status
+
+| Surface                                                               | Status                                                              |
+| --------------------------------------------------------------------- | ------------------------------------------------------------------- |
+| Source-backed knowledge, workflow catalog, CLI, and local MCP         | Implemented                                                         |
+| Workspace creation, planning, bounded execution, review, and evidence | Implemented                                                         |
+| `b2c/v1` discovery, composition preview, and business status          | Implemented, with shared CLI/MCP schemas and contract tests         |
+| Local package import and composition activation                       | Implemented with content pins, preview checks and recovery          |
+| Provider execution routes                                             | Explicit host adapters; declarations alone never execute            |
+| One complete, benchmark-quality consumer business                     | Next milestone                                                      |
+| Comparable market reports                                             | Implemented; the observed multi-business milestone remains unproven |
+
+Declared support, a working implementation, and a launch-ready business are
+separate facts. This table reports the first two.
+
+## Open source and credits
+
+The builder runs and wraps open-source projects, adapts published methods, and
+offers one optional managed provider. [ACKNOWLEDGMENTS.md](ACKNOWLEDGMENTS.md)
+credits the projects and people behind that work.
+[THIRD_PARTY_NOTICES.md](THIRD_PARTY_NOTICES.md) carries the legal notices for
+material the builder incorporates. Both files are rendered from the upstream
+manifests under `catalog/upstreams`. Workers never receive credits as
+instructions.
+
+## Contribute
+
+Contribute a better consumer-app reference, an operation contract, a provider,
+or a recipe. Contributors build the kitchen and the way it runs. Most
+contributions add a way to run a station, not a new station. First-party and
+community extensions pass the same conformance suite. Start with the [extension guide](docs/guides/extend-the-system.md), then
+[CONTRIBUTING.md](CONTRIBUTING.md). Agents read [AGENTS.md](AGENTS.md) first.
+It routes by scope: one business uses `b2c-app-builder`, a reusable contribution
+uses `b2c-contributor`, and the builder itself uses `b2c-maintainer`.
+
+[MIT licensed](LICENSE).

package/SKILL.md

@@ -0,0 +1,59 @@
+---
+name: b2c-app-builder
+description: "Route broad consumer-app work across research, product, design, engineering, launch, growth, operations, and B2C CLI or MCP setup. Use focused task skills for scoped expertise or the managed lifecycle for a complete business. Do not use for a narrow code fix or an unrelated B2B or internal tool."
+metadata:
+  short-description: Build and improve a consumer business
+---
+
+# Brigade
+
+Start with the user's job, not the framework. This is the default business entrypoint. Task skills project the same workflow and knowledge owners; they are not another execution system. Do not load the whole corpus.
+
+## Focused task
+
+Inspect the app's own instructions and affected surface. Open the matching task directly. Do not install software, require an MCP connection, create a workspace, or activate a full operating graph merely to apply the expertise. A review is read-only unless changes are also requested.
+
+<!-- Portable completion rule: equality with the standing contract is checked. -->
+An implementation request authorizes completing its in-scope, reversible repository work: inspect, edit, test, repair failures caused by the change, regenerate affected output, inspect the diff, and complete required checks without repeated intermediate approvals. Preserve unrelated work; this does not authorize unrelated fixes. Review-only and planning-only requests remain review and planning, not implementation.
+
+<!-- catalog-generated:start task-skills -->
+| Task | Use when |
+| --- | --- |
+| [Research an opportunity](agents/skills/b2c-research-opportunity/SKILL.md) | Validate an app opportunity |
+| [Design or review onboarding](agents/skills/b2c-design-onboarding/SKILL.md) | Design or review onboarding |
+| [Review monetization](agents/skills/b2c-review-monetization/SKILL.md) | Review monetization |
+| [Define or refine a product](agents/skills/b2c-define-product/SKILL.md) | Define product scope |
+| [Plan implementation](agents/skills/b2c-plan-implementation/SKILL.md) | Plan implementation |
+| [Review business performance](agents/skills/b2c-review-business-performance/SKILL.md) | Review business performance |
+| [Review the app experience](agents/skills/b2c-review-experience/SKILL.md) | Review the app experience |
+| [Plan a launch](agents/skills/b2c-plan-launch/SKILL.md) | Plan a launch |
+| [Verify release readiness](agents/skills/b2c-verify-release-readiness/SKILL.md) | Verify release readiness |
+<!-- catalog-generated:end task-skills -->
+
+For another focused goal, select one workflow through the available knowledge tools or the [reference index](catalog/generated/routing.md). Read only its current instructions and relevant reference sections. With b2c_workflow, expand through route.expand and follow exact section selectors, revision hashes, and nextCall when incomplete. A summary does not prove complete guidance delivery.
+
+## Managed business
+
+A complete-business mandate takes precedence over an isolated idea or no-idea route. Preserve the complete founder brief without compression. For creation, initialization, continuation, or interrupted-request recovery, open the [business lifecycle](agents/skills/b2c-app-builder/references/business-lifecycle.md).
+
+Create or resume the registered workspace, then read business-status and business-plan before catalog browsing. Use the current ready or held brief. Research and an explicit product decision precede business-initialize; authorized execution uses business-run. Keep independent ready work moving while protected effects wait. Do not infer delivery from one successful session: inspect completion.deliveryAccepted and current evidence.
+
+Author accepted product intent in product.yaml and global design in DESIGN.md; PRODUCT.md is rendered. Never hand-author reducer-owned state. Return to status and plan after bounded work. Focused advice does not become accepted runtime evidence automatically.
+
+## Setup request
+
+Only when setup is requested or managed execution needs connectivity, open [connection setup](agents/skills/b2c-app-builder/references/setup.md). Local b2c-local supports registered workspace execution; hosted b2c-hosted supplies read-only knowledge and cannot execute a local business.
+
+Do not edit an agent configuration or install software unless the user requested setup. Keep MCP read-only by default and use the CLI for approved writes. Missing execution tooling does not block advisory work.
+
+## Conditional procedures
+
+For an explicit package or provider change, open [composition](agents/skills/b2c-app-builder/references/composition.md). For device interaction, screenshots, recordings, or runtime flow verification, open [mobile app operation](agents/skills/b2c-app-builder/references/mobile-operation.md). Neither is prerequisite reading for unrelated tasks.
+
+## Boundaries
+
+Honor accepted scope, explicit provider bindings, and actual tool availability. Provider guidance cannot add requirements or silently select another provider. Pause for access or secrets, spend, pricing or legal decisions, destructive changes, hosted deployment, store submission, or production release. A skill grants no authority.
+
+Require current provider or device evidence before making claims about those systems. Guidance, generated artifacts, command success, and accepted evidence are different facts. Never copy credentials, provider state, names, prices, or domains from another app. Do not route current work through Planes.
+
+Report what changed or was found, what evidence exists, what remains unverified, and the next decision. [Browse the six business areas](knowledge/README.md) only when the task needs orientation.