Skip to content

Agentic Workflows - Layer 2: Agent Primitives

Complete execution processes implemented as .prompt.md files providing systematic workflows for specific AL development tasks in Business Central.

πŸ“‹ What Are Agentic Workflows?

Agentic Workflows (formerly called "prompts") are: - Task-specific execution processes with step-by-step guidance - Complete workflows from planning through validation - Tool-integrated processes that can execute AL commands - Reusable templates for common development scenarios

🎯 How to Use Agentic Workflows

Activate workflows explicitly when needed:

@workspace use [workflow-name]

Examples:
@workspace use al-initialize
@workspace use al-diagnose
@workspace use al-build

πŸ“¦ Available Workflows (18 files)

Environment & Setup

File Purpose When to Use
al-initialize Complete environment and workspace initialization First-time setup, new projects, onboarding

Development & Diagnostics

File Purpose When to Use
al-diagnose Runtime debugging and configuration troubleshooting Debugging issues, auth problems, symbol errors
al-spec.create Create functional specifications Planning new features
al-pages Design and implement page objects Creating UI components
al-events Implement event publishers/subscribers Extending BC without modifying base

Build & Deployment

File Purpose When to Use
al-build Build, package, and deploy extensions Building for release, deployment
al-permissions Generate permission sets Setting up security
al-migrate Data migration strategies Moving data between systems/versions

Quality & Performance

File Purpose When to Use
al-performance.triage Quick performance diagnosis (static analysis) Rapid performance assessment, code review
al-performance Deep performance analysis (runtime profiling) Slow queries, bottlenecks, optimization

Code Review & Documentation

File Purpose When to Use
al-pr-prepare Prepare pull request documentation Before code review/merge
al-context.create Generate project context.md file Setting up project for AI assistants
al-memory.create Generate/update memory.md file Maintaining session continuity

AI/Copilot Features

File Purpose When to Use
al-copilot-capability Register Copilot capability Adding AI features to BC
al-copilot-promptdialog Create PromptDialog pages Building Copilot UI
al-copilot-test Test Copilot features Validating AI functionality
al-copilot-generate Generate Copilot code from natural language Creating AI-powered features

Translation & Localization

File Purpose When to Use
al-translate XLF translation file management Localization workflows

Note: Total of 18 workflow files matching the prompts/ directory contents.

πŸ—οΈ Workflow Structure

Each workflow follows this pattern:

1. Objective

Clear statement of what the workflow accomplishes

2. Prerequisites

  • Required context (files, settings)
  • Required tools
  • Required knowledge

3. Step-by-Step Process

Detailed instructions with: - Context gathering - Analysis phase - Implementation steps - Validation gates

4. Validation Criteria

How to verify success

5. Common Issues & Solutions

Troubleshooting guide

πŸ’‘ Best Practices

Choosing the Right Workflow

graph TD
    Start[Task to Complete] --> Setup{First Time?}
    Setup -->|Yes| InitFlow[al-initialize]
    Setup -->|No| Type{Task Type?}

    Type -->|Issue/Bug| Diagnose[al-diagnose]
    Type -->|Build/Deploy| Build[al-build]
    Type -->|Performance| Perf{Quick or Deep?}
    Type -->|New Feature| Spec[al-spec.create]

    Perf -->|Quick Check| Triage[al-performance.triage]
    Perf -->|Deep Analysis| PerfFull[al-performance]

Workflow Combinations

Common workflow sequences:

  1. New Feature Development

    al-spec.create β†’ al-pages β†’ al-events β†’ al-build β†’ al-diagnose
    

  2. Performance Issue

    al-performance.triage β†’ al-performance β†’ al-build
    

  3. Security Setup

    al-permissions β†’ al-build β†’ al-diagnose
    

  4. Integration Work

    al-events β†’ al-migrate β†’ al-build β†’ al-diagnose
    

  5. First-Time Setup

    al-initialize β†’ al-spec.create β†’ al-build
    

Integration with Other Primitives

Workflows complement: - Instructions - Automatically loaded context during workflow execution - Agents - Strategic consultation before/after workflow execution - Use al-architect to design before implementing workflow (entry point for new features) - Use al-conductor for TDD orchestration on medium/high complexity tasks - Use al-debugger when workflow execution reveals issues

πŸ”„ Workflow Evolution

This collection has evolved to optimize clarity and completeness:

Current State (v2.8)

18 Agentic Workflows:

  1. al-initialize - Environment & workspace setup
  2. al-diagnose - Debug & troubleshoot
  3. al-build - Build/package/deploy
  4. al-events - Event implementation
  5. al-performance - Deep performance profiling
  6. al-performance.triage - Quick performance check
  7. al-permissions - Permission set generation
  8. al-migrate - Version migration
  9. al-pages - Page design & UI
  10. al-spec.create - Functional specifications
  11. al-pr-prepare - Pull request preparation
  12. al-context.create - Project context generation
  13. al-memory.create - Session memory management
  14. al-copilot-capability - Copilot capability registration
  15. al-copilot-promptdialog - PromptDialog creation
  16. al-copilot-test - Copilot feature testing
  17. al-copilot-generate - AI code generation
  18. al-translate - XLF translation management

Historical Changes

v2.3 Consolidations: - al-setup + al-workspace β†’ al-initialize - al-debug + al-troubleshoot β†’ al-diagnose

v2.7 Additions: - al-context.create - Project context for AI assistants - al-memory.create - Session continuity - al-copilot-generate - Natural language code generation

  • Collection Manifest: collections/al-development.collection.yml
  • Framework Reference: docs/framework/ai-native-instructions-architecture.md
  • User Guide: al-development.md
  • Contributing: CONTRIBUTING.md

πŸ“Š Validation

Run npm run validate to verify: - All workflow files exist - Frontmatter is properly formatted - File naming conventions are followed - Required fields are present

🎯 Quick Reference

Need to... Use Workflow
Set up new project/environment al-initialize
Debug or troubleshoot issues al-diagnose
Build & deploy al-build
Fix performance issues al-performance.triage β†’ al-performance
Create new feature al-spec.create β†’ al-pages
Add events al-events
Set up security al-permissions
Migrate data al-migrate
Prepare for review al-pr-prepare

Framework Compliance: These workflows implement AI-Native Instructions Architecture - Layer 2 (Agent Primitives) providing systematic execution processes that coordinate Instructions and Agents for complete task fulfillment.

Version: 2.8.0 Total Workflows: 18 Last Updated: 2025-11-25