Rabetbase CLI — Vibe Coding Dev Toolkit

In one sentence

Rabetbase CLI is the Lovrabet platform development toolkit. Instead of clicking through the Lovrabet workbench, developers complete 95% of the work by natural language right inside the CLI/IDE they already use — Codex, Cursor, Claude Code, Trae, and more. This includes fetching dataset details, developing functions, writing SQL, building pages, incremental DB analysis, generating/modifying standard pages, and more.

How to get the AI started

Goal	Prompt
Query a dataset	"I'm taking over an order management page. Map out the order-related data structures, fields, and relations first, then tell me how to integrate the frontend."
Create a SQL query	"I need a monthly order-amount report. Check whether an existing capability can be reused first; if not, design the SQL and validate the fields and risks."
Push BFF scripts	"Check whether local and remote BFF scripts in this project are in sync. I want to add a customer-dashboard endpoint — give me a plan and an impact preview first, and push only after I confirm."
Troubleshoot the environment	"Check whether this project is bound to a Lovrabet app and environment and whether authentication is still valid. If anything is wrong, list fixes ordered by impact."
Sync menus	"Sync this page's menu to the platform. Preview which menus, resources, and live entry points will change first, and execute only after I confirm."

Recent entry-point reminders

• To find a platform app, just ask the AI to list available apps and tell you which one the current project is bound to.
• To switch the app the current project is bound to, have the AI confirm the project config, target app, and impact scope before rebinding.
• If older material mentions an app entry point that has been retired, let the AI pick the new entry point based on the current Skill and project config — don't copy old commands.

What the AI does when it calls Rabetbase CLI

When you state a need in natural language, the AI breaks it down into verifiable development actions.

1. Understand the goal: identify whether you're doing page integration, data queries, SQL, BFF, menus, troubleshooting, or deployment.
2. Read the real context: use the CLI to fetch the current project, app, environment, dataset, API, SQL, or BFF state — no guessing.
3. Pick the implementation path: prefer the standard SDK and existing platform capabilities; fall back to custom SQL or BFF only when those are genuinely not enough.
4. Preview the risk: read-only actions run immediately; write, push, and delete actions show you the impact scope and dry-run results first.
5. Execute and verify: after you confirm, the AI runs the write and reports the result, failure reasons, and suggested next steps.

What capabilities it provides

The value of Rabetbase isn't "one more CLI" — it lets the AI reliably plug into the real context of the Lovrabet platform, turning development work from console clicking and copy-paste into a trackable, verifiable, automatable workflow.

Dimension	Without the Rabetbase system	With Rabetbase + AI
Data and API alignment	You hunt for datasets, fields, and API contracts by hand; schema changes are easy to miss.	The AI reads the platform's real structure, confirms fields, relations, and available operations, then proposes an integration plan.
Toolchain and context	You bounce between IDE, console, SQL tools, and API testers, carrying context by copy-paste.	The AI pulls context, analyzes the problem, generates a plan, and verifies results — all within one task.
Custom SQL	Fields are guessed, SQL is scattered across environments, and issues surface at integration or after release.	The AI first checks for reusable SQL, then designs the query against real fields, validating and flagging risks before execution.
BFF and local debugging	Scripts only get verified after being pushed, so feedback loops are long.	The AI checks local vs. remote state first, shows the diff, impact scope, and preview, then waits for your confirmation.
Safety and confirmation	You must remember which actions are dangerous; accidental deletes, pushes, or environment changes happen.	The AI reads the CLI's risk levels: read-only actions complete automatically; high-risk writes always explain their impact and wait for confirmation.

Prompts you can copy as-is

These aren't commands — they're requirement descriptions you can send directly to the AI. The AI decides which CLI capabilities to invoke.

Scenario	Example prompt
Taking over a new page	"I'm building an order management page. Map out the order-related datasets, fields, relations, and available operations first, then advise on frontend integration. Don't guess fields — read the real structure."
Writing an aggregate query	"I need monthly order totals. Check whether a standard filter or existing SQL can do it; if not, design a custom SQL and run field validation and risk checks before pushing."
Developing BFF	"I'm adding a customer-dashboard endpoint. Check existing BFF scripts for something similar to avoid duplicate work, then give me an implementation plan, local file changes, and a push preview."
Debugging an issue	"The data on this page looks wrong. Trace it layer by layer — app config, dataset fields, API definitions, SQL/BFF state — and list the evidence at each step."
Pre-release check	"I'm about to release these changes. Verify local and platform state are consistent, list every API, SQL, BFF, or menu affected, and show me the dry-run results first."
Team asset inventory	"Inventory the datasets, SQL, and BFF assets in this app. Flag duplicated capabilities, inconsistent naming, long-unmaintained items, and reusable parts."

Documentation map

This knowledge base is organized into 5 sections. The root page explains the mental model; the hands-on track for project development, page building, configuration, Backend Functions, SQL, and troubleshooting lives in the sub-pages.

Section	Read it when
Getting started	You want to continue with project structure, data models, page development, local debugging, build and packaging, or platform integration from a new project.
Core development capabilities	You're working with APIs, datasets, BFF, custom SQL, code generation, or menu sync.
Maintenance, security, and reference	You need troubleshooting, config references, risk levels, safety confirmations, or command details.
Migration, mapping, and resources	You're migrating from an older CLI, mapping console capabilities, or looking for FAQ, changelog, or screencast material.
Topic Q&A	You're handling customer questions and specific scenarios, such as dataset relations or page buttons calling external APIs.

What to say on your first run

If your environment isn't set up yet, just send this to the AI:

"Check whether this project has a working Rabetbase development environment. If not, install the CLI, log in, initialize the project, and bind the app and environment. If anything involves authorization or high-risk writes, explain why first and wait for my confirmation."