agentleFS
javiarmestojaviarmesto/ALDC-AL-Development-Collection/CLAUDE.md

ALDC-AL-Development-Collection

CLAUDE.md115 starsChanged 10 days agoMITView on GitHub

@import ./docs/framework/ALDC-Core-Spec-v1.2.md AL (Application Language) workspace for Microsoft Dynamics 365 Business Central. Architecture: ALDC Core v1.2 — 13 agents (5 core + 3 on-demand + 4 subagents + 1 extension) + 16 skills + 11 workflows + 9 instructions. 11 workflows (6 core + 5 BC Agents pack). The 6 core, invoked via /workflow-name: 16 composable knowledge modules (8 required + 4 recommended + 3 BC Agents + 1 utility) loaded on-demand by agents (not invoked directly): Agents MUST declare…

The file

We split this CLAUDE.md into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No npm/yarn build steps. AL compilation is handled by the AL Language VS Code extension.

    line 142

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

ALDC — Claude Code Instructions

@import ./docs/framework/ALDC-Core-Spec-v1.2.md

Overview

AL (Application Language) workspace for Microsoft Dynamics 365 Business Central. Architecture: ALDC Core v1.2 — 13 agents (5 core + 3 on-demand + 4 subagents + 1 extension) + 16 skills + 11 workflows + 9 instructions.

Core Principles

  • Extension-only development — Never modify base application objects. Use tableextensions, pageextensions, event subscribers.
  • Human-in-the-Loop (HITL) — All critical decisions require user confirmation before proceeding.
  • TDD / spec-driven — Features follow: architecture (MEDIUM/HIGH) -> al-spec-agent via al-spec.create -> human approval -> test-plan -> implementation -> review.
  • Least privilege — Generate only the minimum permissions required. Use XLIFF for all user-facing strings.

Agent Routing

IntentAgentWhat it does
Specification from approved designdelegate to agent al-spec-agentTechnical contracts and human approval before implementation
Design, architecture, strategydelegate to agent al-architectSolution design, data modeling, integration strategy
Implement, code, debug, fixdelegate to agent al-developerTactical implementation with full AL MCP tools
TDD orchestration (plan -> implement -> review -> commit)delegate to agent al-conductorOrchestrates planning, implementation, and review subagents
Estimate, size, proposedelegate to agent al-presalesPERT estimation, SWOT analysis, cost breakdown

Quick Routing

New feature (MEDIUM/HIGH)? -> al-architect -> /al-spec.create -> al-conductor
New feature (LOW)?         -> /al-spec.create -> al-developer
Bug fix / debugging?       -> al-developer
Architecture review?       -> al-architect
Full TDD cycle?            -> al-conductor
Project estimation?        -> al-presales

Workflows

11 workflows (6 core + 5 BC Agents pack). The 6 core, invoked via /workflow-name:

WorkflowWhen to use
/al-spec.createCreate functional-technical specifications before development
/al-buildBuild, package, and deploy extensions
/al-pr-preparePrepare pull requests with documentation and validation
/al-memory.createGenerate/update memory.md for session continuity
/al-context.createGenerate project context.md for AI assistants
/al-initializeComplete environment and workspace setup

Skills

16 composable knowledge modules (8 required + 4 recommended + 3 BC Agents + 1 utility) loaded on-demand by agents (not invoked directly):

SkillDomainLoaded by
/skill-debugDebugging, snapshot debuggingal-developer
/skill-apiAPI pages, OData, RESTal-developer, al-architect
/skill-copilotAI features, PromptDialogal-developer, al-architect
/skill-eventsEvent subscribers, publishersal-developer, al-architect
/skill-permissionsPermission sets, XLIFF, securityal-developer
/skill-pagesPage types, FastTabs, actionsal-developer
/skill-migrateBC version migration, upgrade codeunitsal-developer
/skill-translateXLF translation, NAB AL Toolsal-developer
/skill-performanceCPU profiling, FlowField optimizational-developer, al-architect
/skill-testingTDD, test strategy, AL Test Toolkital-architect, al-conductor
/skill-estimationPERT estimation, complexity scoringal-presales

Skills Evidencing

Agents MUST declare which skills they loaded and which patterns they applied:

  • al-architect -> > **Skills applied**: skill-api, skill-events at top of architecture.md
  • al-developer -> > **Skills loaded**: skill-debug (root cause analysis) at start of response
  • al-conductor -> Skills Applied in This Phase table in phase-complete.md

Auto-Applied Instructions

Active automatically based on file type (no invocation needed):

Always on *.al: al-guidelines, al-code-style, al-naming-conventions, al-performance Context-activated: al-error-handling (errors), al-events (events), al-testing (test files)

See instructions/ directory for full content.

Plans

Requirement sets live in .github/plans/, one subdirectory per requirement:

.github/plans/
  memory.md                          # Global memory (cross-session decisions)
  {req_name}/
    {req_name}.spec.md               # Functional-technical specification
    {req_name}.architecture.md       # Architecture decisions
    {req_name}.test-plan.md          # Test plan with acceptance criteria
    {req_name}-phase-<N>-complete.md # Phase completion reports
    {req_name}-complete.md           # Final completion report

Workflow with Plans

MEDIUM / HIGH complexity:

  1. al-architect designs solution, creates {req_name}.architecture.md
  2. /al-spec.create reads architecture, generates {req_name}.spec.md
  3. al-conductor reads spec + architecture, orchestrates TDD
  4. /al-pr-prepare prepares PR referencing the plan

LOW complexity:

  1. /al-spec.create generates spec directly from codebase
  2. al-developer implements using spec as blueprint

Complexity-Based Routing

LevelScopeRoute
LOWSingle phase, no integrations/al-spec.create -> al-developer
MEDIUM2-3 areas, internal integrationsal-architect -> /al-spec.create -> al-conductor
HIGH4+ phases, external integrationsal-architect -> /al-spec.create -> al-conductor

Present the assessment and wait for user confirmation before proceeding.

BC Agents Pack (Extension)

For agent development with AI Development Toolkit. Includes al-agent-builder agent and skill-agent-task-patterns (8 SDK integration patterns). Workflows: /al-agent.create, /al-agent.task, /al-agent.instructions, /al-agent.test.

Build & Test

# Download symbols (requires app.json configured)
al-language: Download Symbols

# Build the extension
al-language: Package

# Publish to sandbox
al-language: Publish without Debugging

# Run tests
al-language: Run Tests

No npm/yarn build steps. AL compilation is handled by the AL Language VS Code extension.

Project Structure

instructions/          # Auto-applied instruction files (9)
agents/                # Agent definitions (5 core + 3 on-demand + 3 subagents + 1 extension)
skills/                # Composable knowledge modules (16 skill directories)
prompts/               # Workflow definitions (11 prompt files)
docs/framework/        # Normative spec (ALDC-Core-Spec-v1.2.md)
docs/templates/        # Immutable templates
.github/plans/         # Requirement sets & memory
src/                   # AL source code
app.json               # Extension manifest

Reference


Framework: ALDC Core v1.2 | Primitives: 13 agents + 16 skills + 11 workflows + 9 instructions

For direct Developer increments, use AL Developer Reviewer in an independent context before human approval. Dredd remains the advisory auditor. Both load skill-al-review-pipeline; Conductor phases keep their own review subagent.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.