# Workflow Architecture > Centralized architecture documentation for the pm-workflow bundle with visual diagrams - Skill: `majiayu000/workflow-architecture` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add majiayu000/workflow-architecture` - Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/workflow-architecture/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Docs & Writing - Author: majiayu000 (https://skillmd.com/u/majiayu000) - Updated: 2026-09-09 - Page: https://skillmd.com/skills/majiayu000/workflow-architecture --- # PM-Workflow Architecture **Role**: Central architecture reference for the pm-workflow bundle. Provides visual documentation of the 7-phase execution model, thin agent pattern, and data layer. ``` ┌─────────────────────────────────────────────────────────────────────────────┐ │ │ │ PM-WORKFLOW ARCHITECTURE │ │ │ │ ┌───────────────────────────────────────────────────────────────────────┐ │ │ │ 7-PHASE EXECUTION MODEL │ │ │ │ │ │ │ │ 1-init → 2-refine → 3-outline → 4-plan → 5-execute │ │ │ │ ↓ │ │ │ │ 6-verify ←────┐ │ │ │ │ ↓ │ │ │ │ │ [findings?] │ │ │ │ │ ↓ ↓ │ │ │ │ │ yes no │ │ │ │ │ ↓ ↓ │ │ │ │ │ create fix 7-finalize │ │ │ │ │ tasks (max 3x) │ │ │ │ │ ↓ ↓ │ │ │ │ │ 5-execute [PR issues?] │ │ │ │ │ (loop) ↓ ↓ │ │ │ │ │ ↑ yes no │ │ │ │ │ │ │ ↓ │ │ │ │ │ │ fix tasks COMPLETE │ │ │ │ │ └────────┴──────────────┘ │ │ │ │ │ │ │ │ Iteration Limits: 6-verify (max 5x) | 7-finalize (max 3x) │ │ │ └───────────────────────────────────────────────────────────────────────┘ │ │ │ │ ┌───────────────────────────────────────────────────────────────────────┐ │ │ │ THIN AGENT PATTERN │ │ │ │ │ │ │ │ Orchestrator ──▶ Agent ──▶ Skill │ │ │ │ │ │ │ │ │ │ │ └──▶ Domain Knowledge │ │ │ │ │ │ │ │ │ └──▶ Context Isolation │ │ │ │ │ │ │ └───────────────────────────────────────────────────────────────────────┘ │ │ │ │ ┌───────────────────────────────────────────────────────────────────────┐ │ │ │ DATA LAYER │ │ │ │ │ │ │ │ manage-references manage-lifecycle manage-tasks manage-solution │ │ │ │ │ │ │ │ │ │ │ │ ▼ ▼ ▼ ▼ │ │ │ │ references.toon status.toon TASK-*.toon solution_outline.md │ │ │ │ │ │ │ └───────────────────────────────────────────────────────────────────────┘ │ │ │ └─────────────────────────────────────────────────────────────────────────────┘ ``` --- ## Quick Navigation | Document | Focus | Key Visuals | |----------|-------|-------------| | [standards/phases.md](standards/phases.md) | 7-phase model | Phase flow, transitions, outputs | | [standards/agents.md](standards/agents.md) | Thin agent pattern | Agent structure, skill invocation | | [standards/data-layer.md](standards/data-layer.md) | manage-* skills | File operations, TOON format | | [standards/skill-loading.md](standards/skill-loading.md) | Two-tier loading | System vs domain skills | | [standards/artifacts.md](standards/artifacts.md) | Plan file formats | references.toon, status.toon, TASK-*.toon | | [standards/task-executor-routing.md](standards/task-executor-routing.md) | Task executor routing | Profile→executor mapping, extensibility | | `pm-workflow:workflow-extension-api` | Extension mechanism | Domain extensions for outline/triage | --- ## Core Principles ``` ┌─────────────────────────────────────────────────────────────────────────────┐ │ │ │ CORE DESIGN PRINCIPLES │ │ │ ├─────────────────────────────────────────────────────────────────────────────┤ │ │ │ 1. DOMAIN-AGNOSTIC WORKFLOW │ │ ════════════════════════ │ │ Workflow skills contain NO domain-specific logic. │ │ Domain knowledge comes from marshal.json at runtime. │ │ │ │ 2. THIN AGENT PATTERN │ │ ═══════════════════ │ │ A single parameterized agent (plan-phase-agent) with different │ │ `phase` parameters results in 5 invocation modes, all sharing │ │ one implementation. Agents are minimal wrappers that: │ │ • Resolve skills from marshal.json │ │ • Load resolved skills │ │ • Delegate to skills for actual work │ │ │ │ 3. SINGLE SOURCE OF TRUTH │ │ ════════════════════════ │ │ Plan files (.toon, .md) are the source of truth. │ │ Skills read/write via manage-* scripts only. │ │ │ │ 4. SCRIPT-BASED FILE ACCESS │ │ ═════════════════════════ │ │ ALL .plan/ file access goes through execute-script.py. │ │ NEVER use Read/Write/Edit on .plan/ files directly. │ │ │ └─────────────────────────────────────────────────────────────────────────────┘ ``` --- ## Component Overview ``` ┌─────────────────────────────────────────────────────────────────────────────┐ │ │ │ COMPONENT HIERARCHY │ │ │ │ ┌─────────────────────────────────────────────────────────────────────┐ │ │ │ COMMANDS (User-facing) │ │ │ │ ══════════════════════ │ │ │ │ /plan-marshall /pr-doctor │ │ │ └─────────────────────────────────────────────────────────────────────┘ │ │ │ │ │ ▼ │ │ ┌─────────────────────────────────────────────────────────────────────┐ │ │ │ AGENTS (Single Parameterized Agent) │ │ │ │ ═══════════════════════════════════ │ │ │ │ plan-phase-agent phase=1-init | 2-refine | 3-outline | 4-plan | 5-execute | 6-verify | 7-finalize │ │ │ (One agent, 7 invocation modes) │ │ │ └─────────────────────────────────────────────────────────────────────┘ │ │ │ │ │ ▼ │ │ ┌─────────────────────────────────────────────────────────────────────┐ │ │ │ WORKFLOW SKILLS (Phase Logic) │ │ │ │ ═════════════════════════════ │ │ │ │ phase-1-init phase-2-refine phase-3-outline phase-4-plan │ │ │ │ phase-5-execute phase-6-verify phase-7-finalize │ │ │ │ task-implementation task-module_testing │ │ │ │ git-workflow pr-workflow │ │ │ └─────────────────────────────────────────────────────────────────────┘ │ │ │ │ │ ▼ │ │ ┌─────────────────────────────────────────────────────────────────────┐ │ │ │ DATA LAYER (manage-* Skills) │ │ │ │ ════════════════════════════ │ │ │ │ manage-references manage-lifecycle manage-tasks │ │ │ │ manage-solution-outline manage-plan-documents │ │ │ │ manage-files manage-references │ │ │ └─────────────────────────────────────────────────────────────────────┘ │ │ │ │ │ ▼ │ │ ┌─────────────────────────────────────────────────────────────────────┐ │ │ │ PLAN FILES (.plan/plans/{plan_id}/) │ │ │ │ ═══════════════════════════════════ │ │ │ │ status.toon request.md references.toon solution_outline.md │ │ │ │ TASK-001.toon TASK-002.toon ... │ │ │ └─────────────────────────────────────────────────────────────────────┘ │ │ │ └─────────────────────────────────────────────────────────────────────────────┘ ``` --- ## Related Skills | Skill | Purpose | |-------|---------| | `pm-workflow:plan-marshall` | Unified user-facing entry point for plan lifecycle | | `pm-workflow:workflow-extension-api` | Extension points for domain customization | | `pm-workflow:phase-1-init` | Init phase implementation | | `pm-workflow:phase-2-refine` | Refine phase implementation | | `pm-workflow:phase-3-outline` | Outline phase implementation | | `pm-workflow:phase-4-plan` | Plan phase implementation | | `pm-workflow:phase-5-execute` | Execute phase implementation | | `pm-workflow:phase-6-verify` | Verify phase implementation | | `pm-workflow:phase-7-finalize` | Finalize phase implementation | | `pm-workflow:task-implementation` | Implementation profile workflow | | `pm-workflow:task-module_testing` | Module testing profile workflow | --- ## Standards Documents Load on-demand based on what aspect of the architecture you need to understand: ```bash # Understanding the 7-phase model Read standards/phases.md # Understanding thin agent pattern Read standards/agents.md # Understanding data layer (manage-* skills) Read standards/data-layer.md # Understanding skill loading Read standards/skill-loading.md # Understanding plan file formats Read standards/artifacts.md # Understanding task executor routing Read standards/task-executor-routing.md ```