| name | bkend-guides |
| classification | C |
| description | bkend.ai operational guides, troubleshooting, and platform comparison.
Covers migration guides, performance optimization, testing strategy,
common error solutions, and FAQ.
Triggers: migration, troubleshoot, FAQ, performance, error handling, comparison,
마이그레이션, 문제해결, 자주 묻는 질문, 성능, 에러,
マイグレーション, トラブルシューティング, FAQ, パフォーマンス,
迁移, 故障排除, 常见问题, 性能, 错误处理,
migracion, solucion de problemas, preguntas frecuentes, rendimiento,
migration, depannage, FAQ, performance, gestion des erreurs,
Migration, Fehlerbehebung, FAQ, Leistung, Fehlerbehandlung,
migrazione, risoluzione problemi, FAQ, prestazioni, gestione errori
Do NOT use for: authentication implementation (use bkend-auth),
database schema design (use bkend-data), security setup (use bkend-security)
|
| user-invocable | true |
| argument-hint | |
| allowed-tools | ["read_file","write_file","replace","glob","grep_search","run_shell_command","web_fetch"] |
| imports | [] |
| agents | {"backend":"bkend-expert"} |
| context | session |
| memory | project |
| pdca-phase | all |
bkend-guides
bkend.ai operational guides, troubleshooting, and platform comparison
1. Platform Comparison
| Feature | bkend.ai | Firebase | Supabase |
|---|
| Database | MongoDB Atlas (NoSQL) | Firestore (NoSQL) / RTDB | PostgreSQL (SQL) |
| Auth | Built-in JWT + Social + MFA | Firebase Auth (full suite) | GoTrue (JWT + Social) |
| MCP Support | Native (first-class) | None | Community plugins |
| Real-time | Planned (WebSocket) | Built-in (Firestore listeners) | Built-in (Postgres changes) |
| File Storage | Integrated | Cloud Storage | S3-compatible |
| Edge Functions | Planned | Cloud Functions | Deno Edge Functions |
| Schema | Schemaless / flexible | Schemaless | Rigid SQL migrations |
| Pricing | Usage-based (API calls) | Spark (free) / Blaze (pay-as-you-go) | Free tier + Pro ($25/mo) |
| Open Source | No | No | Yes |
| AI Integration | Native MCP + prompt workflows | Vertex AI extensions | pgvector + AI plugins |
| Best For | AI-assisted rapid prototyping | Mobile-first apps, Google ecosystem | SQL-heavy apps, open source |
When to Choose bkend.ai
- You want AI-assisted development with Gemini CLI, Claude Code, or Cursor via MCP
- You prefer schemaless / flexible data models without migration headaches
- You need rapid prototyping with minimal configuration
- Your team is comfortable with REST APIs and does not need GraphQL
When to Choose Alternatives
- Firebase: You need mature real-time capabilities, deep Google Cloud integration, or are building primarily for mobile within the Google ecosystem
- Supabase: You need SQL, full-text search, row-level security with PostgreSQL policies, or prefer an open-source solution
2. Migration Guides
2.1 From Firebase (Planned)
This migration guide is under development. Key considerations:
- Firestore to bkend.ai: Export Firestore collections as JSON, then import into bkend tables via the REST API or MCP tools
- Firebase Auth to bkend Auth: Re-register users with email/password or social providers; there is no automatic migration path for password hashes
- Cloud Functions to bkend: Rewrite triggers as webhook handlers (webhooks are planned)
- Cloud Storage to bkend Storage: Re-upload files using the bkend storage API
2.2 From Supabase (Planned)
This migration guide is under development. Key considerations:
- PostgreSQL to bkend.ai: Export tables as JSON; flatten relational data into document-style schemas
- Supabase Auth to bkend Auth: Re-register users; JWT tokens are not transferable
- Edge Functions to bkend: Rewrite as API route handlers in your frontend framework
- Supabase Storage to bkend Storage: Re-upload files using the bkend storage API
- RLS Policies: Translate PostgreSQL RLS policies to bkend RBAC rules (admin/user/self/guest)
3. Performance Optimization
3.1 Use Indexes on Frequently Queried Fields
Create indexes on fields that appear in filter and sort parameters. Common candidates:
authorId -- for user-scoped queries
status -- for filtering by state
createdAt -- for chronological sorting
email -- for user lookups
Use the bkend Console or MCP tool to create indexes:
> Create an index on the "posts" table for the "authorId" field
> Create a compound index on "posts" for "status" and "createdAt"
3.2 Limit Query Results
Always set a limit parameter. The maximum allowed value is 100 records per request.
GET /v1/data/posts?limit=20
If no limit is specified, the default is 20. Never rely on fetching all records at once.
3.3 Use Cursor Pagination (Not Offset)
Offset-based pagination (skip) degrades performance on large datasets because the database must scan and discard skipped records.
Avoid (offset pagination):
GET /v1/data/posts?skip=1000&limit=20
Prefer (cursor pagination):
GET /v1/data/posts?cursor=<last-record-id>&limit=20
Cursor pagination is O(1) regardless of page depth.
3.4 Cache with TanStack Query
Configure appropriate staleTime and gcTime for your data:
const queryClient = new QueryClient({
defaultOptions: {
queries: {
staleTime: 5 * 60 * 1000,
gcTime: 30 * 60 * 1000,
retry: 2,
refetchOnWindowFocus: false,
},
},
});
For frequently changing data (like feeds), use shorter staleTime:
useQuery({
queryKey: ["feed"],
queryFn: fetchFeed,
staleTime: 30 * 1000,
});
3.5 Denormalize Frequently Joined Data
bkend.ai uses MongoDB (NoSQL), so joins are not native. Instead of querying multiple tables:
Avoid:
// 1. Fetch post
// 2. Fetch author by post.authorId
// 3. Fetch comments by post._id
// 4. Fetch each comment's author
Prefer: Embed frequently accessed data at write time:
{
"title": "My Post",
"content": "...",
"author": {
"id": "user_123",
"name": "Alice",
"avatar": "https://..."
},
"commentsCount": 5,
"likesCount": 12
}
Update embedded data when the source changes (e.g., when a user updates their name).
4. Testing Strategy
4.1 Unit Tests: Business Logic with Mock bkendFetch
Test business logic in isolation by mocking the API client.
import { canTransition } from "@/application/services/order-state-machine";
describe("Order State Machine", () => {
it("allows draft -> pending", () => {
expect(canTransition("draft", "pending")).toBe(true);
});
it("blocks draft -> shipped", () => {
expect(canTransition("draft", "shipped")).toBe(false);
});
it("allows paid -> cancelled", () => {
expect(canTransition("paid", "cancelled")).toBe(true);
});
it("blocks completed -> any", () => {
expect(canTransition("completed", "pending")).toBe(false);
expect(canTransition("completed", "cancelled")).toBe(false);
});
});
4.2 Integration Tests: API Calls with Dev Environment
Test actual API interactions against the dev environment.
import { bkendFetch } from "@/infrastructure/api/client";
describe("Posts API", () => {
let postId: string;
it("creates a post", async () => {
const res = await bkendFetch("/v1/data/posts", {
method: "POST",
body: JSON.stringify({
title: "Test Post",
content: "Integration test content",
authorId: "test_user_id",
status: "draft",
}),
});
expect(res.success).toBe(true);
postId = res.data._id;
});
it("reads the created post", async () => {
const res = await bkendFetch(`/v1/data/posts/${postId}`);
expect(res.data.title).toBe("Test Post");
});
afterAll(async () => {
if (postId) {
await bkendFetch(`/v1/data/posts/${postId}`, { method: "DELETE" });
}
});
});
4.3 E2E Tests: Full Flow with Test Data
Test complete user flows from login to feature completion.
import { test, expect } from "@playwright/test";
test("create and publish a blog post", async ({ page }) => {
await page.goto("/login");
await page.fill('[name="email"]', "test@example.com");
await page.fill('[name="password"]', "testpassword123");
await page.click('button[type="submit"]');
await expect(page).toHaveURL("/dashboard");
await page.click('a[href="/posts/new"]');
await page.fill('[name="title"]', "E2E Test Post");
await page.fill('[name="content"]', "This is an E2E test.");
await page.click('button:has-text("Publish")');
await page.goto("/posts");
await expect(page.locator("text=E2E Test Post")).toBeVisible();
});
5. Webhooks (Planned)
Webhooks are under development for a future bkend.ai release.
5.1 Table Change Notifications (Planned)
Receive HTTP callbacks when records are created, updated, or deleted in a table.
{
"event": "table.record.created",
"table": "orders",
"record": { "_id": "order_123", "status": "pending", "..." : "..." },
"timestamp": "2026-01-15T10:30:00Z",
"projectId": "proj_abc123",
"environment": "prod"
}
5.2 Auth Event Webhooks (Planned)
Receive callbacks for authentication events.
{
"event": "auth.user.registered",
"user": { "_id": "user_456", "email": "new@example.com" },
"timestamp": "2026-01-15T10:30:00Z",
"projectId": "proj_abc123",
"environment": "prod"
}
Planned auth events: auth.user.registered, auth.user.login, auth.user.logout, auth.user.deleted, auth.token.refreshed.
6. Realtime (Planned)
Real-time subscriptions are under development for a future bkend.ai release.
6.1 WebSocket Subscriptions (Planned)
Subscribe to live data changes via WebSocket connections.
7. Error Handling Guide
7.1 Error Response Format
All bkend.ai API errors follow a consistent JSON structure:
{
"success": false,
"error": {
"code": "VALIDATION_ERROR",
"message": "Invalid email format",
"details": [
{
"field": "email",
"message": "Must be a valid email address"
}
]
}
}
7.2 Common Errors
| Error Code | HTTP Status | Description | Resolution |