Feat/generator dependency registry

.gitignore

@@ -21,3 +21,6 @@ log
 .history
 
 .DS_Store
+
+# Leaked app config from running generators at the repo root during dev/specs
+/config/

README.md

@@ -40,7 +40,14 @@ And then execute:
 
 ## Usage
 
-Run all generators (useful on a new app)
+Run the core generators (recommended on a new app — sets up our standard
+baseline so the app is ready to push straight to Heroku)
+
+```shell
+bin/rails g rolemodel:core_setup
+```
+
+Or run every generator, including app-specific extras like React, SaaS/Devise, and GoodJob
 
 ```shell
 bin/rails g rolemodel:all
@@ -64,10 +71,37 @@ You can see complete list of available generators (including those under the Rol
 bin/rails g
 ```
 
+## Generator Registry
+
+Every rolemodel generator that is run records itself in your app's
+`config/initializers/rolemodel_generators.rb`. This registry powers coupling
+declarations between generators — for example, the sentry and webpack generators
+know about each other through the registry and automatically wire the Sentry
+webpack plugin when both are present.
+
+For apps set up before the registry existed, run the seeding generator:
+
+```shell
+bin/rails g rolemodel:registry
+```
+
+The seeder probes for each generator's characteristic output files and writes
+entries marked as `seeded-by-detection`. It is safe to run multiple times.
+
+### Opting out
+
+* **Persistent opt-out:** Set `g.rolemodel <key>: false` in the initializer.
+  The generator will never re-record over a `false` entry.
+* **Per-invocation opt-out:** Pass `--no-<key>` to any generator that declares
+  a coupling (e.g. `rails g rolemodel:sentry --no-sentry-webpack`).
+* **Drift recovery:** If the managed block markers are missing from the
+  initializer, delete the file and run `rails g rolemodel:registry` to
+  rebuild it.
+
 ## Generators
 
+* [Core Setup](./lib/generators/rolemodel/core_setup)
 * [Github](./lib/generators/rolemodel/github)
-* [Semaphore](./lib/generators/rolemodel/semaphore)
 * [Heroku](./lib/generators/rolemodel/heroku)
 * [Readme](./lib/generators/rolemodel/readme)
 * [Webpack](./lib/generators/rolemodel/webpack)
@@ -101,10 +135,11 @@ bin/rails g
 * [Editors](./lib/generators/rolemodel/editors)
 * [Tailored Select](./lib/generators/rolemodel/tailored_select)
 * [Lograge](./lib/generators/rolemodel/lograge)
+* [Registry](./lib/generators/rolemodel/registry)
 
 ## Development
 
-Install the versions of Node and Ruby specified in `.node-version` and `.ruby-version` on your machine. https://asdf-vm.com/ is a great tool for managing language versions. Then run `npm install -g yarn`.
+Install the versions of Node and Ruby specified in `.node-version` and `.ruby-version` on your machine. https://asdf-vm.com/ is a great tool for managing language versions. Then run `corepack enable` to activate the Yarn 4+ version pinned by each project's `packageManager` field.
 
 ## Adding new Generators
 

lib/generators/rolemodel/README.md

@@ -13,13 +13,32 @@
 * [Optics](./optics)
 * [README](./readme)
 * [SaaS](./saas)
-* [Semaphore](./semaphore)
 * [SimpleForm](./simple_form)
 * [Slim](./slim)
 * [SoftDestroyable](./soft_destroyable)
 * [Source Map](./source_map)
 * [Testing](./testing)
 * [Webpack](./webpack)
+* [Registry](./registry)
+
+## Generator Registry
+
+Each generator records itself in `config/initializers/rolemodel_generators.rb`
+so that other generators can detect whether it has been applied. See
+[Registry](./registry) for the seeding tool.
+
+### Generator coupling
+
+Some generators declare optional couplings:
+
+* **sentry ↔ webpack:** Whichever is installed second wires the Sentry webpack
+  plugin into `webpack.config.js` via the `sentry_webpack` hook sub-generator.
+  Pass `--no-sentry-webpack` to suppress. Set
+  `g.rolemodel sentry_webpack: false` in the initializer for a persistent opt-out.
+* **tailored_select ↔ simple_form:** Installing simple_form with
+  `--tailored-select` installs the tailored_select package and its input.
+  Installing tailored_select standalone installs the package without the input.
+  Pass `--simple-form-input` to force the input regardless.
 
 ## Helpful documentation
 

lib/generators/rolemodel/all_generator.rb

@@ -2,12 +2,16 @@ module Rolemodel
   class AllGenerator < GeneratorBase
     source_root File.expand_path('templates', __dir__)
 
+    # Composite generator: it orchestrates other generators and installs nothing
+    # itself, so it is not recorded in the registry.
+    skip_registry_entry!
+
     def run_all_the_generators
       generate 'rolemodel:github'
-      generate 'rolemodel:semaphore'
       generate 'rolemodel:heroku'
       generate 'rolemodel:readme'
       generate 'rolemodel:webpack'
+      generate 'rolemodel:sentry'
       generate 'rolemodel:react'
       generate 'rolemodel:slim'
       generate 'rolemodel:optics:all'

lib/generators/rolemodel/core_setup/README.md

@@ -0,0 +1,27 @@
+# Core Setup Generator
+
+Runs the core generators every new Rails app should have. Unlike
+`rolemodel:all`, this skips the app-specific extras (React, SaaS/Devise,
+GoodJob, Kaminari, etc.) so you get just the standard baseline — a new Rails
+app you can push straight to Heroku right after generation.
+
+## Prerequisites
+
+  - A freshly generated Rails app
+
+## What you get
+
+  - [GitHub](../github) — standard GitHub configuration
+  - [Heroku](../heroku) — standard Heroku deployment configuration
+  - [Readme](../readme) — standard project README
+  - [Webpack](../webpack) — Webpack v5 for JS and CSS
+  - [Sentry](../sentry) — error monitoring for Ruby and JavaScript
+  - [Slim](../slim) — Slim templates
+  - [Optics](../optics) — Optics design system
+  - [Testing](../testing) — RSpec, FactoryBot, parallel_tests, TestProf
+    (pass `--js-runner` to include jasmine-playwright-runner)
+  - [SimpleForm](../simple_form) — SimpleForm with our configuration
+  - [Linters](../linters) — Rubocop and ESLint
+  - [UI Components](../ui_components) — flash, the modal pattern, & Turbo 8 support
+  - [Editors](../editors) — EditorConfig and recommended VSCode extensions
+  - [Lograge](../lograge) — condensed request logging

lib/generators/rolemodel/core_setup/USAGE

@@ -0,0 +1,12 @@
+Description:
+  Runs the core generators every new Rails app should have
+
+  Sets up GitHub config, Heroku deployment, README, Webpack, Sentry, Slim,
+  Optics, testing (RSpec & friends), SimpleForm, linters, Turbo 8+ support,
+  and Lograge. The result is an app that is ready to push directly to Heroku.
+
+  Pass --js-runner to also include jasmine-playwright-runner for browser
+  JS testing.
+
+Example:
+  rails generate rolemodel:core_setup

lib/generators/rolemodel/core_setup/core_setup_generator.rb

@@ -0,0 +1,23 @@
+module Rolemodel
+  class CoreSetupGenerator < ::Rolemodel::GeneratorBase
+    # Composite generator: it orchestrates other generators and installs nothing
+    # itself, so it is not recorded in the registry.
+    skip_registry_entry!
+
+    def run_the_core_generators
+      generate 'rolemodel:github'
+      generate 'rolemodel:heroku'
+      generate 'rolemodel:readme'
+      generate 'rolemodel:webpack'
+      generate 'rolemodel:sentry'
+      generate 'rolemodel:slim'
+      generate 'rolemodel:optics:all'
+      generate 'rolemodel:testing:all'
+      generate 'rolemodel:simple_form'
+      generate 'rolemodel:linters:all'
+      generate 'rolemodel:ui_components:flash'
+      generate 'rolemodel:ui_components:modals'
+      generate 'rolemodel:lograge'
+    end
+  end
+end

lib/generators/rolemodel/github/README.md

@@ -9,6 +9,10 @@ It doesn't need to be run first, but the parallel_tests generator must be run in
 - CI workflow
   - A sensible default `ci.yml` to get you started with Github Actions. This will run linters, model tests, and system tests.
   - Along with the `ci.yml`, your `database.yml` will be modified to be able to be run in GHA.
+- Deploy workflows
+  - `deploy-staging.yml` deploys to Heroku on every push to `main` (or manually); `deploy-production.yml` deploys manually via `workflow_dispatch`.
+  - Both target a GitHub deployment environment (`Staging`/`Production`) that provides `HEROKU_APP_NAME` and `HEROKU_APP_URL` variables, and authenticate with the org-level `HEROKU_IT_SUPPORT_API_KEY` secret and `HEROKU_IT_SUPPORT_EMAIL` variable.
+  - Note: the staging workflow will fail on pushes to `main` until the environment exists — run the `deploy-app` agent skill (installed by the heroku generator) to create the Heroku app and the GitHub environment.
 - Pull Request Template
   - When you open a Pull Request in Github it will use the Markdown file as a [template](./templates/pull_request_template.md) for the content of the PR.
   - Helpful for reminding collaborators to add specific details to the PR.

lib/generators/rolemodel/github/templates/workflows/deploy-production.yml

@@ -0,0 +1,34 @@
+name: Deploy to Heroku Production
+
+on:
+  workflow_dispatch:
+    inputs:
+      sha:
+        description: 'Specific SHA or Branch name (optional)'
+        required: false
+        type: string
+
+concurrency: production-deployment
+
+jobs:
+  deploy:
+    name: Deploy
+    runs-on: ubuntu-latest
+    timeout-minutes: 20
+    environment:
+      name: Production
+      url: ${{ vars.HEROKU_APP_URL }}
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: Install Heroku CLI
+        run: |
+          curl https://cli-assets.heroku.com/install.sh | sh
+
+      - uses: akhileshns/heroku-deploy@v3.15.15
+        with:
+          heroku_api_key: ${{ secrets.HEROKU_IT_SUPPORT_API_KEY }}
+          heroku_app_name: ${{ vars.HEROKU_APP_NAME }}
+          heroku_email: ${{ vars.HEROKU_IT_SUPPORT_EMAIL }}
+          branch: ${{ inputs.sha || 'HEAD' }}
+          healthcheck: '${{ vars.HEROKU_APP_URL }}/up'

lib/generators/rolemodel/github/templates/workflows/deploy-staging.yml

@@ -0,0 +1,35 @@
+name: Deploy to Heroku Staging
+
+on:
+  workflow_dispatch:
+    inputs:
+      sha:
+        description: 'Specific SHA or Branch name (optional)'
+        required: false
+        type: string
+  push: { branches: [ staging ] }
+
+concurrency: staging-deployment
+
+jobs:
+  deploy:
+    name: Deploy
+    runs-on: ubuntu-latest
+    timeout-minutes: 20
+    environment:
+      name: Staging
+      url: ${{ vars.HEROKU_APP_URL }}
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: Install Heroku CLI
+        run: |
+          curl https://cli-assets.heroku.com/install.sh | sh
+
+      - uses: akhileshns/heroku-deploy@v3.15.15
+        with:
+          heroku_api_key: ${{ secrets.HEROKU_IT_SUPPORT_API_KEY }}
+          heroku_app_name: ${{ vars.HEROKU_APP_NAME }}
+          heroku_email: ${{ vars.HEROKU_IT_SUPPORT_EMAIL }}
+          branch: ${{ inputs.sha || 'HEAD' }}
+          healthcheck: "${{ vars.HEROKU_APP_URL }}/up"

lib/generators/rolemodel/good_job/good_job_generator.rb

@@ -64,8 +64,6 @@ def copy_initializers
 
     def finishing_notes
       say <<~NOTES
-        *** Reminder to update Honeybadger gem to version 5.7.0 or later to get correct GoodJob error notifications in Honeybadger
-
         *** Reminder to also update your job classes to include appropriate concurrency controls (enqueue_limit/perform_limit with keys)
       NOTES
     end

lib/generators/rolemodel/heroku/README.md

@@ -4,5 +4,15 @@
 
 * Procfile configured for a server process and a release command to run migrations
 * Basic app.json preconfigured with a script to initialize the database and common environment variables, addons, and buildpacks
+* A pointer in `AGENTS.md` to the `deploy-app` agent skill, which ships inside the
+  `rolemodel-rails` gem (`lib/rolemodel/skills/deploy-app/SKILL.md`) rather than being copied
+  into your repo — it's one-time deployment setup, so nothing skill-related is committed to the
+  app. The skill cleans up the generated Gemfile (merges duplicate groups, removes comments,
+  alphabetizes), verifies the test suite and RuboCop pass, creates the Sentry project and wires
+  up the DSN, creates and deploys the Heroku staging app (buildpacks, dynos, Postgres,
+  Papertrail), and creates the GitHub `Staging` environment with the
+  `HEROKU_APP_NAME`/`HEROKU_APP_URL` variables the deploy workflow consumes. The skill is
+  LLM-agnostic (Agent Skills format): locate it with `bundle show rolemodel-rails` and point any
+  coding agent at the SKILL.md. Requires the Heroku CLI, the GitHub CLI, and the Sentry MCP server.
 
 This is the basic config needed to deploy to Heroku.

lib/generators/rolemodel/heroku/heroku_generator.rb

@@ -15,6 +15,47 @@ def install_procfile
       template 'Procfile'
     end
 
+    def reference_deploy_app_skill
+      say 'Reference the deploy-app agent skill from rolemodel-rails', :green
+
+      # The deploy-app skill is one-time, run-once deployment setup. Rather than
+      # copying ~200 lines of instructions into every app repo, point the agent at
+      # the SKILL.md that ships inside the rolemodel-rails gem (already a bundled
+      # dependency). Nothing skill-related is committed to the generated app.
+      if File.exist?(File.join(destination_root, 'AGENTS.md'))
+        append_to_file 'AGENTS.md', agents_md_skill_entry
+      else
+        create_file 'AGENTS.md', "# Agent instructions\n#{agents_md_skill_entry}"
+      end
+    end
+
+    def pin_ruby_version_for_buildpack
+      say 'Pin the Ruby version in the Gemfile so the Heroku buildpack respects it.', :green
+
+      # A bare .ruby-version file is not enough: without a `ruby` directive the version
+      # never lands in Gemfile.lock, so the Heroku Ruby buildpack falls back to its own
+      # default and can install an incompatible Ruby. Tie the Gemfile to .ruby-version.
+      gemfile = File.join(destination_root, 'Gemfile')
+      return if File.exist?(gemfile) && File.read(gemfile).match?(/^\s*ruby\s/)
+
+      inject_into_file 'Gemfile', "\nruby file: '.ruby-version'\n", after: /^source .*$/
+    end
+
+    def use_database_url_in_production
+      say 'Point the production database at DATABASE_URL for Heroku.', :green
+
+      # Rails' generated production block hardcodes database/username/<APP>_DATABASE_PASSWORD,
+      # none of which exist on Heroku, where the Postgres add-on provides a full DATABASE_URL.
+      # Replace the whole block with a url-based config.
+      gsub_file 'config/database.yml',
+                /^production:\n(?:[ \t]+.*\n?)+/,
+                <<~YAML
+                  production:
+                    <<: *default
+                    url: <%= ENV["DATABASE_URL"] %>
+                YAML
+    end
+
     def force_ssl
       say 'Require SSL for production environment.', :green
 
@@ -48,5 +89,26 @@ def create_assets_rake_tasks # rubocop:disable Metrics/MethodLength
         end
       RAKE
     end
+
+    private
+
+    def agents_md_skill_entry
+      <<~MD
+
+        ## Agent skills
+
+        Reusable, agent-agnostic task instructions (Agent Skills format) ship inside the
+        `rolemodel-rails` gem — a bundled dependency of this app — rather than being copied
+        into this repo. Locate the gem's skills directory with
+        `bundle show rolemodel-rails` (the skills live under `lib/rolemodel/skills/`), then
+        read the relevant `SKILL.md` and follow it directly.
+
+        * `deploy-app` (`$(bundle show rolemodel-rails)/lib/rolemodel/skills/deploy-app/SKILL.md`)
+          — one-time deployment setup: cleans up the generated Gemfile, verifies the test
+          suite and RuboCop pass, creates the Sentry project and wires up the DSN, creates
+          and deploys the Heroku staging app, and creates the GitHub `Staging` environment +
+          deploy workflow. Use when asked to set up staging, Heroku, or deployment.
+      MD
+    end
   end
 end

lib/generators/rolemodel/heroku/templates/app.json.tt

@@ -4,12 +4,15 @@
     "postdeploy": "bin/rails db:seed"
   },
   "env": {
-    "HONEYBADGER_API_KEY": {
+    "SENTRY_DSN": {
       "required": true
     },
-    "HONEYBADGER_ENV": {
+    "SENTRY_ENVIRONMENT": {
       "required": true,
       "value": "review-app"
+    },
+    "SENTRY_AUTH_TOKEN": {
+      "required": false
     }
   },
   "addons": [