backend-manager 5.9.5 → 5.9.7

package/plans/mcp2.md

@@ -1,247 +0,0 @@
-# MCP Role-Based Tool Scoping + Consumer Extensibility
-
-## Context
-
-The BEM MCP server currently works only for admins — a single `BACKEND_MANAGER_KEY` grants access to all 19 tools. Regular users can't connect at all, and consumer BEM projects have no way to add custom MCP tools. This plan adds:
-
-1. **Role-based tool scoping** — tools tagged `admin`, `user`, or `public`; connections only see tools matching their role
-2. **User authentication** — seamless OAuth sign-in flow + API key as long-lived credential
-3. **Consumer MCP tools** — a single `functions/mcp.js` file in consumer projects
-
-## Architecture
-
-### Role Model
-
-Every tool gets a `role` field:
-
-| Role | Who sees it | Examples |
-|------|-------------|---------|
-| `admin` | Admin key connections only | firestore_read, send_email, run_cron |
-| `user` | Authenticated users + admins | get_user, get_subscription, cancel_subscription |
-| `public` | Everyone (including unauthenticated) | generate_uuid, health_check |
-
-Admin sees ALL tools. User sees `user` + `public`. Unauthenticated sees only `public`. Defense-in-depth — even if someone calls an admin tool by name, the route still rejects with 403.
-
-### Auth: How Users Connect (OAuth + Consumer's Website)
-
-**The key insight:** users already have an API key (`api.privateKey`) on their Firestore account doc. It's long-lived, already works with `assistant.authenticate()`, and already encodes identity + access level. This is the credential.
-
-**The UX:** Claude Code/Desktop opens a browser → user signs in on their familiar site → redirected back to Claude. Done. No copy-pasting tokens.
-
-**The OAuth flow (step by step):**
-
-```
-1. Claude discovers endpoints:
-   GET /.well-known/oauth-authorization-server
-   → { authorization_endpoint, token_endpoint, ... }
-
-2. Claude opens browser to BEM authorize endpoint:
-   GET /backend-manager/mcp/authorize?redirect_uri=CLAUDE_CALLBACK&state=STATE
-
-3. BEM authorize checks the request:
-   - If client_id === BACKEND_MANAGER_KEY → auto-redirect (admin, unchanged)
-   - Otherwise → redirect to consumer's /token page:
-     https://app.example.com/token?redirect_uri=CLAUDE_CALLBACK&state=STATE
-
-4. User lands on their familiar website sign-in page.
-   Signs in with Google / email+password / whatever.
-   After sign-in, page gets Firebase ID token via webManager.auth().getIdToken()
-
-5. Consumer's /token page redirects to CLAUDE_CALLBACK:
-   CLAUDE_CALLBACK?code=FIREBASE_ID_TOKEN&state=STATE
-
-6. Claude exchanges code for access token:
-   POST /backend-manager/mcp/token
-   { code: FIREBASE_ID_TOKEN }
-
-7. BEM token endpoint:
-   - If code === BACKEND_MANAGER_KEY → return it as access_token (admin, unchanged)
-   - Otherwise → verify Firebase ID token with admin.auth().verifyIdToken()
-     → look up user doc → return user's api.privateKey as access_token
-
-8. Claude uses the API key for all future MCP requests:
-   Authorization: Bearer {api.privateKey}
-```
-
-**Why API key, not JWT:**
-- Firebase ID tokens expire in 1 hour — MCP connection would break constantly
-- The API key never expires (until regenerated)
-- `assistant.authenticate()` already handles API key → user lookup
-- User can revoke by regenerating from account settings
-
-**Why redirect to consumer's website instead of embedding Firebase Auth:**
-- User sees their familiar sign-in page (branded, trusted)
-- No need to embed Firebase Auth SDK in BEM's authorize HTML
-- UJM already has a `/token` page — just needs a small update to support `redirect_uri` param
-- Works with any auth provider the consumer has configured
-
-### Auth Classification (Fast, No DB Call)
-
-`resolveAuthInfo(token)` classifies the Bearer token for tool filtering:
-
-- Token matches `BACKEND_MANAGER_KEY` → `role: 'admin'`
-- Any other non-empty token → `role: 'user'` (API key from the OAuth flow)
-- No token → `role: 'public'`
-
-No DB call needed — actual validation happens at the route level when a tool is called.
-
-### Consumer MCP Tools
-
-Consumer tools live in a **single file**: `functions/mcp.js`. Keeps things simple since most consumer MCP tools are just route delegations pointing at routes already defined in `functions/routes/`.
-
-```js
-// functions/mcp.js
-module.exports = [
-  // Route delegation — points at an existing route (works on stdio + HTTP)
-  {
-    name: 'get_sponsorship',
-    description: 'Get sponsorship details by ID',
-    role: 'user',
-    method: 'GET',
-    path: 'sponsorship',
-    inputSchema: {
-      type: 'object',
-      properties: {
-        id: { type: 'string', description: 'Sponsorship ID' },
-      },
-      required: ['id'],
-    },
-  },
-
-  // Handler mode — runs code directly (HTTP transport only, has full Manager context)
-  {
-    name: 'newsletter_stats',
-    description: 'Get newsletter stats for the past N days',
-    role: 'admin',
-    inputSchema: {
-      type: 'object',
-      properties: {
-        days: { type: 'number', description: 'Days to look back', default: 30 },
-      },
-    },
-    handler: async ({ Manager, assistant, user, params, libraries }) => {
-      const cutoff = Date.now() - (params.days || 30) * 86400000;
-      const snapshot = await libraries.admin.firestore()
-        .collection('newsletters')
-        .where('metadata.created.timestampUNIX', '>=', Math.floor(cutoff / 1000))
-        .get();
-      return { total: snapshot.docs.length };
-    },
-  },
-];
-```
-
-Consumer tools with the same name as a built-in tool override it (same precedence as consumer routes).
-
-## File Changes
-
-### 1. `src/mcp/tools.js` — Add `role` to every tool
-
-| Tool | Role |
-|------|------|
-| `firestore_read/write/query` | `admin` |
-| `send_email`, `send_notification` | `admin` |
-| `sync_users`, `list_campaigns`, `create_campaign` | `admin` |
-| `get_stats`, `run_cron`, `create_post`, `create_backup`, `run_hook` | `admin` |
-| `get_user`, `get_subscription` | `user` |
-| `cancel_subscription`, `refund_payment` | `user` |
-| `generate_uuid`, `health_check` | `public` |
-
-### 2. NEW: `src/mcp/utils.js` — Shared utilities
-
-- **`resolveAuthInfo(token)`** — classifies token → `{ role, authType, token }`. Admin key check is instant; everything else = user.
-- **`filterToolsByRole(tools, role)`** — admin→all, user→user+public, public→public only.
-- **`loadConsumerTools(cwd)`** — checks for `${cwd}/mcp.js`, `require()`s it if it exists, validates shape, returns array.
-- **`buildToolMap(builtinTools, consumerTools)`** — merges into a Map; consumer tools override same-name built-ins.
-
-### 3. `src/mcp/client.js` — Support user auth tokens
-
-Extend constructor to accept `{ baseUrl, backendManagerKey, userToken }`.
-
-In `call()`:
-- Has `backendManagerKey` → current behavior (key in query/body)
-- Has `userToken` (no backendManagerKey) → put token in `Authorization: Bearer` header + `authenticationToken` query param for GET
-- Neither → unauthenticated request
-
-### 4. `src/mcp/index.js` — Stdio server with role filtering
-
-- Import utils, call `resolveAuthInfo()` to determine role
-- Call `loadConsumerTools(options.cwd)` if `cwd` is provided
-- Merge + filter tools by role in `ListToolsRequestSchema` handler
-- In `CallToolRequestSchema`: `path`-based tools via BEMClient. Handler-only tools return error explaining they require HTTP transport.
-- Accept new `options.token` for user connections
-
-### 5. `src/mcp/handler.js` — HTTP handler with OAuth user flow
-
-**`handleAuthorize()`** — the big change:
-- If `client_id` matches admin key → auto-redirect (current behavior, unchanged)
-- Otherwise → **redirect to consumer's website** for sign-in:
-  - Build consumer auth URL from `Manager.config.brand.url` (or a configurable `mcp.authUrl` in backend-manager-config.json)
-  - Redirect to: `{consumerUrl}/token?redirect_uri={originalRedirectUri}&state={state}`
-  - The consumer's `/token` page handles sign-in, gets Firebase ID token, redirects back to Claude's callback
-
-**`handleToken()`** — exchanges Firebase ID token for API key:
-- If code matches admin key → return as `access_token` (unchanged)
-- Otherwise → treat code as Firebase ID token:
-  1. `admin.auth().verifyIdToken(code)` to get UID
-  2. `admin.firestore().doc('users/{uid}').get()` to get user doc
-  3. Return `user.api.privateKey` as the `access_token`
-  4. If user has no API key, generate one and save it
-
-**`handleMcpProtocol()`** — role-based tool filtering:
-- Extract Bearer token, call `resolveAuthInfo()`
-- Allow public connections (no 401 for empty tokens — just show public tools)
-- Discover + merge consumer tools (cached at module scope)
-- Filter tools by role in `ListToolsRequestSchema`
-- In `CallToolRequestSchema`:
-  - `path`-based tools: BEMClient HTTP call (current behavior)
-  - `handler`-based tools: execute directly with `{ Manager, assistant, user, params, libraries }`
-
-**Rename `isValidKey()` → `isAdminKey()`** for clarity.
-
-### 6. `src/cli/commands/mcp.js` — New CLI flags
-
-- `--token` (`-t`): user's API key (for user-level connections)
-- Pass `cwd: functionsDir` to `startServer()` for consumer tool discovery
-- When `--token` is provided, create BEMClient with userToken instead of backendManagerKey
-
-### 7. Consumer's UJM `/token` page — Small update (separate repo)
-
-The UJM `/token` page needs to support the MCP OAuth redirect flow:
-- Detect `redirect_uri` and `state` query params
-- After sign-in, get Firebase ID token via `webManager.auth().getIdToken()`
-- Redirect to `redirect_uri?code={idToken}&state={state}`
-
-This is a small addition to the existing `/token` page layout. Files:
-- `ultimate-jekyll-manager/src/defaults/dist/_layouts/blueprint/auth/token.html`
-- `ultimate-jekyll-manager/src/defaults/dist/_layouts/themes/classy/frontend/pages/auth/token.html`
-
-### 8. `docs/mcp.md` — Documentation
-
-- Add "Roles" section explaining admin/user/public scoping
-- Add "User Authentication" section with the OAuth flow diagram
-- Add "Consumer MCP Tools" section with `functions/mcp.js` format + examples
-- Add CLI examples for user connections
-- Add guide for configuring the consumer auth URL
-
-## Implementation Order
-
-1. `src/mcp/utils.js` (new) — foundation utilities
-2. `src/mcp/tools.js` — add `role` to all 19 tools
-3. `src/mcp/client.js` — user token support
-4. `src/mcp/index.js` — stdio role filtering + consumer tools
-5. `src/cli/commands/mcp.js` — new CLI flags
-6. `src/mcp/handler.js` — HTTP role filtering + OAuth user flow + consumer tool handlers
-7. UJM `/token` page update (separate repo)
-8. `docs/mcp.md` — documentation
-
-## Verification
-
-1. **Existing admin flow**: `npx mgr mcp` with `BACKEND_MANAGER_KEY` → all 19 tools listed, all callable
-2. **User flow (stdio)**: `npx mgr mcp --token <api-key>` → only user+public tools listed
-3. **Public flow**: `npx mgr mcp` (no key, no token) → only public tools listed
-4. **Consumer tools**: Create `functions/mcp.js` in a consumer project → tools appear in listing
-5. **HTTP OAuth flow**: Connect from Claude Code/Desktop → redirected to consumer site → sign in → redirected back → user tools available
-6. **Token exchange**: POST to `/mcp/token` with Firebase ID token → returns API key as access_token
-7. **Role enforcement**: User calling an admin tool by name → unknown tool error (tool not in filtered list)
-8. **Run `npx mgr test`** to verify no regressions

package/plans/suspended-subscription-dead-zone.md

@@ -1,97 +0,0 @@
-# Suspended Subscription Dead Zone
-
-## Issue
-
-A user with a `suspended` subscription was trapped in a loop:
-- **Trying to subscribe** → "You already have a subscription. Please cancel your existing subscription before purchasing a new one." (400)
-- **Trying to cancel** → "No active paid subscription found" (400)
-
-No way out.
-
-## Affected User
-
-- **UID:** `t9AeAe7QUhNXAUYRV1vUbOU0QVV2`
-- **Brand:** Somiibo
-
-The user had two stale subscriptions from different processors:
-
-| Field | Processor | Status | Resource ID | Last Webhook |
-|---|---|---|---|---|
-| `subscription` | PayPal | `suspended` | `I-Y41DMAGNWGP1` | `BILLING.SUBSCRIPTION.SUSPENDED` (2026-04-19) |
-| `plan` (legacy) | Chargebee | `suspended` | `169xhEU1pYXAt3dsj` | `subscription-profile-fixer` (2024-02-28) |
-
-PayPal suspended the subscription (likely failed payment) but never sent a `BILLING.SUBSCRIPTION.CANCELLED` webhook. PayPal doesn't auto-cancel suspended subscriptions — it leaves them in limbo until the user resolves payment or manually cancels in PayPal. The local status stayed `suspended` forever.
-
-## Root Cause
-
-Asymmetric status gates between the intent and cancel endpoints:
-
-### Intent gate (`POST /payments/intent`)
-```js
-// Blocked if product is NOT basic AND status is NOT cancelled
-if (subProductId !== 'basic' && subStatus !== 'cancelled') { ... }
-```
-- `active` → blocked (correct)
-- `suspended` → blocked (correct — user should cancel first)
-- `cancelled` → allowed (correct)
-
-### Cancel gate (`POST /payments/cancel`)
-```js
-// Blocked if status is NOT active
-if (subscription.status !== 'active') { ... }
-```
-- `active` → allowed (correct)
-- `suspended` → **blocked** (the bug)
-- `cancelled` → blocked (correct — nothing to cancel)
-
-The intent gate treats `suspended` as "has a subscription" (must cancel first), but the cancel gate treats `suspended` as "no subscription to cancel." Dead zone.
-
-## The Fix
-
-### 1. Cancel gate now accepts `suspended`
-
-```js
-// Before
-if (!subscription || subscription.status !== 'active' || ...)
-
-// After
-if (!subscription || (subscription.status !== 'active' && subscription.status !== 'suspended') || ...)
-```
-
-### 2. Fallback for dead processor subscriptions
-
-When cancelling a `suspended` subscription, the processor API call might fail — the subscription could be expired/deleted on the processor's side while our local state is stale. The cancel endpoint now catches processor errors for suspended subscriptions and directly resets the user doc:
-
-```js
-if (subscription.status === 'suspended') {
-  // Processor rejected — subscription is already dead on their end
-  // Directly reset the user doc to cancelled
-  await admin.firestore().doc(`users/${uid}`).set({
-    subscription: {
-      status: 'cancelled',
-      product: { id: 'basic', name: 'Basic' },
-      cancellation: { pending: false, date: { timestamp, timestampUNIX } },
-    },
-  }, { merge: true });
-}
-```
-
-### 3. New test coverage
-
-- Test account `cancel-suspended` with `status: 'suspended'` and test processor payment details
-- Test `allows-suspended-subscription` verifying the cancel endpoint accepts and processes suspended subscriptions
-
-## Status Matrix (after fix)
-
-| Status | Can subscribe? | Can cancel? | Can delete account? |
-|---|---|---|---|
-| `active` (basic) | Yes | No (nothing to cancel) | Yes |
-| `active` (paid) | No (cancel first) | Yes | No |
-| `suspended` | No (cancel first) | **Yes** (fixed) | No |
-| `cancelled` | Yes | No (already cancelled) | Yes |
-
-## Files Changed
-
-- `src/manager/routes/payments/cancel/post.js` — gate + fallback
-- `src/test/test-accounts.js` — `cancel-suspended` account
-- `test/routes/payments/cancel.js` — `allows-suspended-subscription` test

package/scripts/test-helper-providers.js

@@ -1,162 +0,0 @@
-#!/usr/bin/env node
-
-/**
- * Test Helper — direct provider lookups + removals via API
- *
- * Lets the live-test checklist verify SendGrid + Beehiiv state without
- * clicking around in dashboards. Reuses the same provider helpers that
- * BEM uses in production (findContact / removeContact).
- *
- * Usage (run from a consumer project's functions/ dir):
- *
- *   node ../../../backend-manager/scripts/test-helper-providers.js find  user@example.com
- *   node ../../../backend-manager/scripts/test-helper-providers.js purge user@example.com
- *
- * Or symlink: `ln -s ../../../backend-manager/scripts/test-helper-providers.js ./prov`
- * then: `node ./prov find user@example.com`
- *
- * - find:  prints whether the contact exists in SendGrid + Beehiiv (and the data shape).
- * - purge: removes the contact from BOTH providers (idempotent; safe if absent).
- *
- * The script picks up:
- *   - SENDGRID_API_KEY, BEEHIIV_API_KEY from <cwd>/.env (functions/.env)
- *   - marketing.campaigns.listId, marketing.newsletter.publicationId from <cwd>/backend-manager-config.json
- *   - service-account from <cwd>/service-account.json
- *
- * Exit codes: 0 ok, 1 usage error, 2 provider error.
- */
-
-const path = require('path');
-const fs = require('fs');
-
-const cwd = process.cwd();
-const envPath = path.join(cwd, '.env');
-const configPath = path.join(cwd, 'backend-manager-config.json');
-const serviceAccountPath = path.join(cwd, 'service-account.json');
-
-// --- sanity checks ---
-if (!fs.existsSync(envPath)) {
-  console.error(`✗ Missing ${envPath} — run from the consumer's functions/ dir`);
-  process.exit(1);
-}
-if (!fs.existsSync(configPath)) {
-  console.error(`✗ Missing ${configPath} — run from the consumer's functions/ dir`);
-  process.exit(1);
-}
-if (!fs.existsSync(serviceAccountPath)) {
-  console.error(`✗ Missing ${serviceAccountPath} — run from the consumer's functions/ dir`);
-  process.exit(1);
-}
-
-// --- parse args ---
-const [, , cmd, email] = process.argv;
-const VALID_CMDS = ['find', 'purge'];
-
-if (!cmd || !VALID_CMDS.includes(cmd) || !email) {
-  console.error('Usage:');
-  console.error('  node test-helper-providers.js find  <email>');
-  console.error('  node test-helper-providers.js purge <email>');
-  process.exit(1);
-}
-
-// --- bootstrap env (dotenv style, minimal) ---
-require('dotenv').config({ path: envPath });
-
-if (!process.env.SENDGRID_API_KEY) {
-  console.error('✗ SENDGRID_API_KEY not set in .env');
-  process.exit(1);
-}
-if (!process.env.BEEHIIV_API_KEY) {
-  console.error('✗ BEEHIIV_API_KEY not set in .env');
-  process.exit(1);
-}
-
-// --- bootstrap Manager so providers can read Manager.config.marketing.* ---
-// The providers require '../../../index.js' which is the Manager singleton.
-// We need to load BEM from the consumer's node_modules (not the BEM repo's own src)
-// so it picks up the consumer's config + service account.
-let Manager;
-try {
-  const BackendManager = require(path.join(cwd, 'node_modules', 'backend-manager'));
-  Manager = (new BackendManager()).init({}, { setupFunctionsLegacy: false, log: false });
-} catch (e) {
-  console.error('✗ Failed to bootstrap Manager from consumer node_modules:', e.message);
-  process.exit(2);
-}
-
-// --- load providers via the BEM Manager.libraries surface (preferred) or direct path ---
-const sendgridProviderPath = path.join(cwd, 'node_modules', 'backend-manager', 'src', 'manager', 'libraries', 'email', 'providers', 'sendgrid.js');
-const beehiivProviderPath = path.join(cwd, 'node_modules', 'backend-manager', 'src', 'manager', 'libraries', 'email', 'providers', 'beehiiv.js');
-
-const sendgrid = require(sendgridProviderPath);
-const beehiiv = require(beehiivProviderPath);
-
-// --- commands ---
-async function find(email) {
-  console.log(`Looking up ${email} on both providers...`);
-
-  const [sgContact, bhContact] = await Promise.all([
-    sendgrid.findContact(email).catch(e => ({ _error: e.message })),
-    beehiiv.findContact(email).catch(e => ({ _error: e.message })),
-  ]);
-
-  console.log('\n── SendGrid ──');
-  if (sgContact?._error) {
-    console.log(`  ✗ error: ${sgContact._error}`);
-  } else if (!sgContact) {
-    console.log('  ⊘ not found');
-  } else {
-    console.log(`  ✓ found — id=${sgContact.id || '?'}, email=${sgContact.email || '?'}`);
-  }
-
-  console.log('\n── Beehiiv ──');
-  if (bhContact?._error) {
-    console.log(`  ✗ error: ${bhContact._error}`);
-  } else if (!bhContact) {
-    console.log('  ⊘ not found');
-  } else {
-    console.log(`  ✓ found — id=${bhContact.id || '?'}, email=${bhContact.email || '?'}, status=${bhContact.status || '?'}`);
-  }
-
-  return { sgContact, bhContact };
-}
-
-async function purge(email) {
-  console.log(`Removing ${email} from both providers...`);
-
-  const [sgResult, bhResult] = await Promise.all([
-    sendgrid.removeContact(email).catch(e => ({ _error: e.message })),
-    beehiiv.removeContact(email).catch(e => ({ _error: e.message })),
-  ]);
-
-  console.log('\n── SendGrid ──');
-  if (sgResult?._error) {
-    console.log(`  ✗ error: ${sgResult._error}`);
-  } else {
-    console.log(`  ✓ ${JSON.stringify(sgResult)}`);
-  }
-
-  console.log('\n── Beehiiv ──');
-  if (bhResult?._error) {
-    console.log(`  ✗ error: ${bhResult._error}`);
-  } else {
-    console.log(`  ✓ ${JSON.stringify(bhResult)}`);
-  }
-
-  return { sgResult, bhResult };
-}
-
-// --- main ---
-(async () => {
-  try {
-    if (cmd === 'find') {
-      await find(email);
-    } else if (cmd === 'purge') {
-      await purge(email);
-    }
-    process.exit(0);
-  } catch (e) {
-    console.error('✗ Unexpected error:', e);
-    process.exit(2);
-  }
-})();

package/scripts/update-disposable-domains.js

@@ -1,44 +0,0 @@
-#!/usr/bin/env node
-
-/**
- * Fetches the latest disposable email domain list from GitHub and saves it as JSON.
- *
- * Source: https://github.com/disposable-email-domains/disposable-email-domains
- * (curated, ~5k domains, low false-positive rate)
- *
- * Run manually:   node scripts/update-disposable-domains.js
- * Runs automatically on: npm prepublishOnly
- */
-const fs = require('fs');
-const path = require('path');
-
-const SOURCE_URL = 'https://raw.githubusercontent.com/disposable-email-domains/disposable-email-domains/main/disposable_email_blocklist.conf';
-const OUTPUT_PATH = path.join(__dirname, '..', 'src', 'manager', 'libraries', 'email', 'data', 'disposable-domains.json');
-
-async function main() {
-  console.log('Fetching disposable domain list...');
-
-  const response = await fetch(SOURCE_URL);
-
-  if (!response.ok) {
-    throw new Error(`Failed to fetch: ${response.status} ${response.statusText}`);
-  }
-
-  const text = await response.text();
-  const domains = text
-    .trim()
-    .split('\n')
-    .map(d => d.trim().toLowerCase())
-    .filter(Boolean);
-
-  const unique = [...new Set(domains)].sort();
-
-  fs.writeFileSync(OUTPUT_PATH, JSON.stringify(unique, null, 2) + '\n');
-
-  console.log(`Updated disposable-domains.json: ${unique.length} domains`);
-}
-
-main().catch((e) => {
-  console.warn('Warning: Failed to update disposable domains:', e.message);
-  console.warn('Using existing list. Run manually later: node scripts/update-disposable-domains.js');
-});

package/templates/_.env

@@ -1,42 +0,0 @@
-# ========== Default Values ==========
-# GitHub
-GH_TOKEN=""
-
-# Backend Manager
-BACKEND_MANAGER_KEY=""
-BACKEND_MANAGER_WEBHOOK_KEY=""
-BACKEND_MANAGER_NAMESPACE=""
-BACKEND_MANAGER_OPENAI_API_KEY=""
-BACKEND_MANAGER_ANTHROPIC_API_KEY=""
-
-# AI
-OPENAI_API_KEY=""
-ANTHROPIC_API_KEY=""
-CLAUDE_CODE_OAUTH_TOKEN=""
-
-# Payment Processors
-PAYPAL_CLIENT_SECRET=""
-STRIPE_SECRET_KEY=""
-CHARGEBEE_API_KEY=""
-COINBASE_API_KEY=""
-
-# Analytics
-GOOGLE_ANALYTICS_SECRET=""
-META_ACCESS_TOKEN=""
-TIKTOK_ACCESS_TOKEN=""
-
-# Security
-CLOUDFLARE_TOKEN=""
-RECAPTCHA_SECRET_KEY=""
-
-# Email & Newsletter
-SENDGRID_API_KEY=""
-BEEHIIV_API_KEY=""
-NEVERBOUNCE_API_KEY=""
-ZEROBOUNCE_API_KEY=""
-UNSUBSCRIBE_HMAC_KEY=""
-APOLLO_API_KEY=""
-
-# ========== Custom Values ==========
-# Add your custom environment variables below this line
-# ...

package/templates/_.gitignore

@@ -1,60 +0,0 @@
-# ========== Default Values ==========
-# macOS
-.DS_Store
-
-# Logs
-logs
-*.log
-npm-debug.log*
-yarn-debug.log*
-yarn-error.log*
-firebase-debug.log*
-
-# Firebase cache
-.firebase/
-
-# Firebase config
-.firebaserc
-
-# Runtime data
-pids
-*.pid
-*.seed
-*.pid.lock
-
-# Coverage directory
-coverage
-.nyc_output
-
-# Dependency directories
-node_modules/
-
-# Optional npm cache directory
-.npm
-
-# Optional eslint cache
-.eslintcache
-
-# Output of 'npm pack'
-*.tgz
-
-# Yarn Integrity file
-.yarn-integrity
-
-# dotenv environment variables file
-.env*
-
-# BEM specific
-.runtimeconfig.json
-service-account*.json
-bem-reload-trigger.js
-.temp/
-_legacy
-
-# Root proxy (generated by setup)
-/package.json
-/package-lock.json
-
-# ========== Custom Values ==========
-# Add your custom ignore patterns below this line
-# ...