Ctrlk

Agent Orchestration Prompts

One tool call to rule them all. Tell your AI agent what you want to do and the agent orchestrates the entire workflow — extraction, JSON generation, UI derivation, and implementation completion — automatically.

How to Use

Just tell your AI agent what you want to do. The agent detects your intent, loads the right conventions and guides, then executes the workflow autonomously.

Copy-paste any prompt below into your AI agent chat:

Phase 1 — Extraction & Analysis

Analyze a BRD:

Analyze this BRD and prepare FlexBase files

Migrate a legacy application:

Migrate this legacy application to FlexBase

Migrate with stored procedure & trigger extraction:

Migrate this legacy application to FlexBase — extract stored procedures and triggers too

Design a new application:

Design a new scheduling app for healthcare

Full pipeline (extract → Studio → complete):

Build me the full app from this requirements doc

Phase 3 — Implementation Completion

Complete API business logic (after FlexBase Studio code generation):

Complete with stored procedure strategy override:

Phase 3 — UI Generation

Generate Blazor UI for a module:

Setup FlexBase theme (one-time, before any UI work):

Standalone — Optional Tasks

Apply original database names to domain models:

Integrate a DataStore or Provider:

Task Catalog

Task 1: Full Pipeline (Extract → Studio Handoff → Complete)

Goal: End-to-end: analyze source → generate JSON → [user does Studio] → complete API + UI User provides: Source material (BRD, legacy code, or business requirements)

Task 2: Extract from Source (Phase 1 Only)

Goal: Analyze source material and produce Markdown + JSON files User provides: BRD, legacy code, or business description Produces: Markdown extraction files, JSON import files with UI metadata, Studio handoff instructions

Task 3: Complete Implementation (Phase 3 Only)

Goal: Complete API business logic and generate UI pages from generated skeleton code User provides: Generated FlexBase solution + extraction artifacts (JSON/Markdown)

Task 4: Resume After Studio

Goal: Pick up where Phase 1 left off — user has generated code and wants implementation completed User provides: Says "I've generated the code" or "complete the implementation"

Task 5: Generate UI Only — Complete Module UI from One Command

Goal: Generate the complete UI stack for a module — REST client, DTOs, dialog, list page, navigation — without touching the API project User provides: Module name (e.g., "Products", "Orders", "Customers") and optionally UI framework preference User says: "Generate Blazor pages for the Products module" or "Build the UI for Orders"

Task 6: Setup FlexBase Theme for Blazor FluentUI (One-Time)

Goal: Configure business-domain-themed admin layout with FluentDesignTheme, sidebar navigation, and FluentUI providers — first step before any functional UI work User provides: Generated Blazor FluentUI solution + business domain context

Task 7: Apply Original Database Names to Domain Models

Goal: Add [Table] and [Column] EF Core annotations to domain models for legacy database compatibility User provides: FlexBase solution with domain models + extraction JSON containing originalName / sourceTable metadata This task is standalone — run it independently at any time after domain models are generated.

Task 8: Legacy Migration with Database Stored Procedure & Trigger Extraction

Goal: Same as Task 2 (Legacy Migration extraction) plus extract stored procedure and trigger definitions from the legacy database User provides: Legacy source material (+ optionally exported SP/trigger scripts) User says: "Migrate this legacy application to FlexBase — extract stored procedures and triggers too" Default behavior: SP logic will be implemented as C# code in Phase 3 User override: Say "keep the stored procedures" to retain them in the database

Task 9: Stored Procedure Implementation Strategy (Phase 3) — OVERRIDE Only

Goal: Override the default SP handling during implementation completion When needed: Only when the user explicitly says "use stored procedures" or "keep stored procedures". The default implementation automatically translates SP logic to C# code without any extra prompt. Options:

  • Move to Code (DEFAULT): SP logic becomes C# handler/query code

  • Call Stored Procedures: Keep SPs in the database and call them from FlexBase handlers

  • Mixed Strategy: Per-feature control — some SPs move to code, others stay as stored procedures

Task 10: DataStore & Provider Integration

Goal: Integrate FlexBase DataStore or Provider operations into handlers, queries, subscribers, or validation plugins User provides: The store/provider type and business use case User says: Any of the following:

Supported stores and providers:

Type
Examples

Document Store

CosmosDB, MongoDB, DynamoDB, RavenDB

Vector Store

PgVector, Azure AI Search, OpenSearch

Graph Store

Neo4j, Apache AGE, Amazon Neptune

Search Store

Elasticsearch, Azure AI Search, Algolia

File Store

Azure Blob, AWS S3, GCP Storage

Time Series Store

TimescaleDB, InfluxDB, Azure Data Explorer

Cache Store

Redis, Memcached, NCache

AI Provider

Azure OpenAI, OpenAI, Anthropic, Gemini, Ollama

Email Provider

SMTP, SendGrid, Console (dev)

Sub-Prompt: Workflow State Extraction

When: During Phase 1 extraction, any domain entity that has a status/state column (e.g., Status, State, OrderStatus, IsActive) Goal: Extract complete workflow state definitions and transitions per entity — handled automatically during extraction.

Intent Detection — Quick Reference

User Says
What Happens

"analyze this BRD", "extract requirements"

Phase 1 — BRD extraction

"migrate this app", "convert this code"

Phase 1 — Legacy migration

"design a new app", "build me an app for..."

Phase 1 — New application

"complete the implementation", "finish the code"

Phase 3 — Implementation completion

"generate Blazor pages", "build the UI"

Phase 3 — UI generation

"setup FlexBase theme"

Phase 3 — Theme setup

"apply original database names"

Phase 3 — Domain name annotations

"migrate with stored procedures"

Phase 1 — Legacy + SP extraction

"use stored procedures", "create database scripts"

Phase 3 — SP implementation override

"add CosmosDB to handler", "set up email notifications"

Phase 3 — DataStore/Provider integration

Scope Control

The agent stays within the documented scope. It does NOT:

  • Invent features not in the source material

  • Add entities not in the extraction JSON

  • Create UI screens for features that don't exist

  • Expand scope unless the user explicitly asks

Last updated