Skip to content

WAY_OF_WORK.md

Latest commit

 

History

History
95 lines (75 loc) · 3.36 KB

WAY_OF_WORK.md

File metadata and controls

95 lines (75 loc) · 3.36 KB
Raw

WAY OF WORK

Overview

This project follows the 7-phase research-first methodology proven in three production skill packages (ERPNext 28 skills, Blender-Bonsai 73 skills, Tauri 2 27 skills). The methodology ensures deterministic, high-quality skills by mandating deep research before any skill creation.

Core principle: You cannot create deterministic skills for something you don't deeply understand.

Full methodology: See https://github.com/OpenAEC-Foundation/Skill-Package-Workflow-Template/blob/main/WORKFLOW.md

The 7 Phases

Phase 1: Raw Masterplan

  • Define project scope and Tailwind CSS coverage areas
  • Create preliminary skill inventory (estimate, not final)
  • Set up repository structure and core files
  • Output: docs/masterplan/{{TECH_PREFIX}}-masterplan.md, all core files

Phase 2: Deep Research (Vooronderzoek)

  • One comprehensive research document for Tailwind CSS
  • Cover: API surface, architecture, patterns, anti-patterns, version differences
  • Minimum 2000 words, verify with WebFetch
  • Output: docs/research/vooronderzoek-{{TECH_PREFIX}}.md

Phase 3: Masterplan Refinement

  • Review research against preliminary inventory
  • Add, merge, or remove skills based on findings
  • Define dependencies and batch execution order
  • Write ready-to-use agent prompts for each skill
  • Output: updated masterplan (definitive)

Phase 4: Topic-Specific Research

  • Per-skill focused research (concurrent with Phase 5)
  • Only the information that specific skill needs
  • Verify against official docs using WebFetch
  • Output: docs/research/topic-research/{skill-name}-research.md

Phase 5: Skill Creation

  • Transform research into deterministic skills
  • Execute in batches of 3 agents via Claude Code Agent tool
  • Quality gate after every batch
  • Output: skills/source/{{TECH_PREFIX}}-{category}/{skill-name}/

Phase 6: Validation

  • Structural, content, cross-reference, and functional validation
  • All skills must pass quality gate before publication
  • Output: validation report

Phase 7: Publication

  • INDEX.md, updated README.md, social preview banner
  • GitHub remote under OpenAEC Foundation
  • Release tag (v1.0.0)

Skill Structure

Directory Layout

skill-name/
├── SKILL.md              # Main file, < 500 lines
└── references/
    ├── methods.md        # Complete API signatures
    ├── examples.md       # Working code examples
    └── anti-patterns.md  # What NOT to do

Naming Convention

  • {{TECH_PREFIX}}-{category}-{topic}
  • Categories: syntax, impl, errors, core, agents

Content Standards

DO:

  • Use deterministic language: "ALWAYS use X when Y" / "NEVER do X because Y"
  • Verify ALL code against official documentation (WebFetch)
  • Version-explicit code examples
  • Document anti-patterns with explanations
  • Keep SKILL.md under 500 lines

DON'T:

  • Vague language: "you might consider" is BANNED
  • Assumptions about API behavior
  • Outdated or unverified code
  • Non-English content
  • Training data without WebFetch verification

Orchestration Model

  • Main session = ORCHESTRATOR (coordinates, validates)
  • Agents = WORKERS (research, write, validate)
  • 3 agents per batch, quality gate between batches
  • Each agent writes to unique directory (no conflicts)

Version Control

  • Commit after EVERY phase: Phase X.Y: [action] [subject]
  • ROADMAP.md updated with every commit
  • Push to GitHub after every phase