Using AI With IDP Projects
Once AI context has been added to your application repository (see Adding AI Context), your AI assistant understands the IDP framework and can assist with concrete development tasks. The following sections show the main ways to work with it effectively.
General Framework Questions
Ask your AI assistant about any framework workflow in plain language. The AI context gives it the knowledge to answer accurately for IDP rather than giving generic answers.
Validation
- "How do I add a uniqueness check to an entity validator?"
- "Where should I add an async database-backed validation rule?"
- "How do I add a rule that only applies on insert?"
Application bootstrap
- "How do I register the main window when creating a new Prism application?"
- "How do I add a shared UI service to the application builder?"
Navigation and modules
- "How do I navigate to a list view from a menu action?"
- "Where do I register a view into a Prism region?"
- "How do I wire back-navigation after a user saves an edit screen?"
Screen and view-model behavior
- "How do I hook into the save flow before an entity is persisted?"
- "How do I restore view state when a user returns to a list screen?"
- "How do I subscribe to an entity event to refresh a list?"
Telerik grids and lists
- "How do I persist column visibility so it survives application restarts?"
- "How do I synchronize the selected grid rows with a view-model collection?"
- "How do I add a custom link-button column to a grid?"
Entity-Specific and Project-Relevant Tasks
The AI assistant is not limited to generic framework questions. Because the AI context includes knowledge of the framework's extension points and safe customization seams, you can ask questions that are specific to the entities and screens in your own application.
Examples:
- "The
Orderentity needs a validation rule that prevents deletion when open invoice lines exist. How do I implement this?" - "Add a validation rule to the
Customerentity that checks the email address is unique across all customers." - "The
Productlist screen should restore the last used sort order when the user navigates back to it. How do I add this?" - "I need to navigate from the
Invoiceedit screen back to theInvoicelist after a successful save. How do I wire this?" - "The
OrderLinegrid should allow the user to show and hide the discount column and remember the choice. How do I add this?"
The assistant will respond with concrete implementation steps that follow the correct IDP patterns for each task.
Business Requirements to Technical Implementation Plan
Before writing any code, you can ask the assistant to translate a business requirement into a structured technical plan. This is useful for scoping a feature, identifying which parts of the framework are involved, and catching missing details early.
Write the requirement in plain business language and ask the assistant to produce an implementation plan. For example:
"A customer must not be able to place a new order if they have any unpaid invoices older than 90 days. The system should show a clear message explaining why the order was blocked."
The assistant can respond with a plan that covers:
- which validator and rule set to add the check to
- whether the check needs an async database call
- where the user-facing error message should be registered
- which screen or command triggers the validation
- the order in which to implement and test the pieces
This pattern works well for any feature that touches validation, navigation, screen behavior, or data access. Once you are happy with the plan, you can work through it step by step and ask the assistant to help with each piece.
A Note on Keeping Answers Current
The skill and agent content that powers the AI assistant's answers is maintained alongside the framework source. It reflects the current implementation rather than being a static snapshot. The learn site does not duplicate this content — ask your assistant directly for the most up-to-date guidance on any workflow.