Modifying Modules

Add fields, states, or activities to existing modules without disrupting live data.

Module modification tools are available in configure mode (the default) and frontend mode.

To check or switch your current mode, tell the AI: "Switch to configure mode."

How It Works

The AI uses the modify_module guided prompt. It retrieves the current module definition with stable internal IDs, applies your requested change, validates it locally, then submits the update.

1

Describe the change

Tell the AI which module to modify and what you want to change.

"Add a Priority field (High/Medium/Low) to the Purchase Order module."

"Add an 'Escalate to Director' activity to the Leave Request module."

"Rename the 'Pending Review' state to 'Under Review' in the Invoice module."

2

AI retrieves the canvas

The AI calls get_module_canvas to fetch the full current module definition, including stable IDs for every field, state, activity, and flow. These IDs are what allow renaming without data loss.

3

Review the proposed changes

The AI shows you exactly what will change — new items added, existing items modified — before committing.

4

Validate and update

The AI calls validate_design with mode=update to check the modified schema locally. Once validation passes, it calls update_module to merge the changes.

Sections you didn't mention are left unchanged.

What You Can Change

Change type	Example
Add a field	"Add a Notes field to the Inspection module"
Rename a field	"Rename 'Remarks' to 'Comments'"
Add a state	"Add a 'Pending Payment' state after Approved"
Change a state color	"Make the Escalated state red"
Add an activity	"Add a 'Request More Info' activity"
Add a flow	"Allow the 'Request More Info' activity from Pending Review"
Change an activity actor	"Make the Approve activity AI-hybrid instead of human-only"

Safety

update_module merges changes into the existing schema using stable IDs. Items not included in the update are preserved. The AI never deletes fields, states, or activities unless you explicitly ask it to.

If you ask the AI to remove something that has existing data (e.g., delete a field that entries already use), it will warn you before proceeding.