Add Dataverse skill

📋 Shared Instructions: shared-instructions.md - Cross-cutting concerns.

References:

• dataverse-reference.md - Picklist fields, virtual fields, lookups, file/image columns, form patterns (CRITICAL)
• api-authentication-reference.md - Dataverse API auth, token, publisher prefix
• table-management-reference.md - Query, create, extend tables and columns
• data-architecture-reference.md - Relationship types, dependency tiers

Add Dataverse

Two paths: existing tables (skip to Step 5) or new tables (full workflow).

Workflow

1. Plan → 2. Setup API Auth → 3. Review Existing Tables → 4. Create Tables → 5. Add Data Source → 6. Review Generated Files → 7. Build

---

Step 1: Plan

Check memory bank for project context. Ask the user:

1. Which Dataverse table(s) do they need? (e.g., account, contact, cr123_customentity)
2. Do the tables already exist in their environment, or do they need to create new ones?

If tables already exist: Skip to Step 5.

If creating new tables:

• Ask about the data they need and design an appropriate schema
• Use standard Dataverse tables when appropriate (contact for people, account for organizations)
• Build a dependency graph -- see data-architecture-reference.md for tier classification
• Enter plan mode with EnterPlanMode, present ER model with tables, columns, relationships, and creation order
• Get approval with ExitPlanMode

Step 2: Setup API Auth (if creating tables)

See api-authentication-reference.md for full details.

az account show   # Verify Azure CLI logged in

# Find your Dataverse environment URL:
# In make.powerapps.com → Settings → Developer resources → Web API endpoint
# It looks like: https://<org-name>.crm.dynamics.com/api/data/v9.2/
# Use the base URL: https://<org-name>.crm.dynamics.com

$api = Initialize-DataverseApi -EnvironmentUrl "https://<org>.crm.dynamics.com"
$headers = $api.Headers
$baseUrl = $api.BaseUrl
$publisherPrefix = $api.PublisherPrefix

Requires System Administrator or System Customizer security role.

Step 3: Review Existing Tables (if creating tables)

Always query existing tables first before creating:

$existingTables = Invoke-RestMethod -Uri "$baseUrl/EntityDefinitions?`$filter=IsCustomEntity eq true&`$select=SchemaName,LogicalName,DisplayName" -Headers $headers

See table-management-reference.md for Find-SimilarTables, Compare-TableSchemas, and Build-TableNameMapping functions.

Present findings to user with AskUserQuestion:

• Tables that can be reused (already exist with matching columns)
• Tables that need extension (exist but missing columns)
• Tables that must be created (no match found)

Step 4: Create Tables (if creating tables)

Get explicit confirmation before creating. Create in dependency order:

• Tier 0: Reference tables (no dependencies)
• Tier 1: Primary entities (reference Tier 0)
• Tier 2: Dependent tables (reference Tier 1)

Use safe functions from table-management-reference.md:

• New-DataverseTableIfNotExists
• Add-DataverseColumnIfNotExists
• Add-DataverseLookupIfNotExists (from data-architecture-reference.md)

Step 5: Add Data Source

For each table:

pa app add data-source --connector dataverse --table <table-logical-name>

Can add multiple tables by running the command for each one.

Step 6: Review Generated Files

The command generates:

• src/generated/models/{Table}Model.ts -- TypeScript interfaces, plus {Table}FileColumnName, {Table}ImageColumnName, {Table}UploadColumnName union types if the table has file/image columns
• src/generated/services/{Table}Service.ts -- CRUD methods (create, get, getAll, update, delete) plus upload, downloadFile, downloadImage, deleteFileOrImage if file/image columns exist

Show the user a usage example:

import { AccountsService } from "../generated/services/AccountsService";

const result = await AccountsService.getAll({
  select: ["name", "accountnumber"],
  filter: "statecode eq 0",
  orderBy: ["name asc"],
  top: 50
});
const accounts = result.data || [];

Key rules:

• Use generated services (e.g., AccountsService.getAll()), not fetch/axios
• Check result.data for actual data
• Don't edit generated files unless needed
• Read dataverse-reference.md before writing any Dataverse code -- picklist fields, virtual fields, lookups, and file/image columns all have critical gotchas

Step 7: Build

npm run build

Fix TypeScript errors before proceeding. Do NOT deploy yet.

Update Memory Bank

Record which tables were added (or created), generated files, and any schema notes.