agentleFS
TheBeardedBearSASTheBeardedBearSAS/claude-craft/.claude/skills/architect/SKILL.md

architect

Skill108 starsChanged 6 days agoMITView on GitHub

Phase d'architecture systématique AVANT le code (TDD). Use when starting a new feature that touches multiple files, designing a new module, or making non-trivial architectural choices.

190 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use ecosystem-reference, fractal-delegation and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • naman10parikhSkill

    ecosystem-reference

    Quick-access directory of Claude Code ecosystem repos, skills, plugins, MCPs, and patterns. Use when improving the harness.

    Also in 5 projects that use this

  • naman10parikhSkill

    fractal-delegation

    Also in 5 projects that use this

  • jaydubya818Skill

    figure-it-out

    Use when novel or multi-phase work needs an auditable workflow and no narrower playbook fits.

    Also in 63 projects that use this

  • bnemaSkill

    setup-pstack

    Configure which Pi models pstack uses per role. Detects available provider/model specs and writes pstack's user configuration. Use for /setup-pstack, "configure pstack models", or changing pstack model choices.

    Also in 32 projects that use this

  • shiwenbin1617Skill

    ketchup-video

    Catch the user up like /ketchup, with a short subtitled video of slides for what happened and the waiting calls as text under it. Use for /ketchup-video.

    Also in 20 projects that use this

  • jaydubya818Skill

    poteto-mode

    Use when non-trivial work should follow the smallest rigorous Jstack workflow.

    Also in 37 projects that use this

  • +5 more from naman10parikh

The file

Files
SKILL.md

Architect — Phase d'architecture avant TDD

Skill inspiré de obra/superpowers. Objectif : produire une spec d'architecture avant d'écrire le moindre test ou ligne de code.

Règle d'or : si tu ne peux pas dessiner le diagramme, tu ne peux pas coder la feature.

Quand utiliser

SituationArchitect phase ?
Bug 1 fichier❌ Non (go direct TDD)
Nouvelle feature < 3 fichiers⚠️ Court (10 min)
Nouvelle feature > 3 fichiers✅ Obligatoire
Nouveau module/bounded context✅ Obligatoire
Migration technique✅ Obligatoire
Intégration externe (API, MQ, DB)✅ Obligatoire

Processus en 5 étapes

1. Identifier les boundaries

Quelles frontières la feature touche-t-elle ?

  • Domain boundaries — quels bounded contexts impliqués ?
  • Layer boundaries — Presentation / Application / Domain / Infrastructure ?
  • Process boundaries — sync HTTP / async queue / event-driven ?
  • Data boundaries — quelles bases de données, quelles tables ?
  • Ownership boundaries — quelle équipe owne quel morceau ?

Output : liste des frontières traversées.

2. Définir les contrats

Pour chaque frontière traversée, définir le contrat :

  • Inputs : format, types, invariants, nullabilité
  • Outputs : format, codes d'erreur, latence cible
  • Side effects : écritures DB, events émis, appels externes
  • Idempotence : le contrat est-il idempotent ?
  • Transactions : atomique ? eventually consistent ?

Output : signatures de fonctions/endpoints + DTO.

3. Dessiner les dépendances

Qui dépend de quoi ? Dans quel sens ?

[HTTP Controller] ──▶ [Use Case] ──▶ [Domain Service]
                                          │
                                          ▼
                                     [Repository] (interface)
                                          │
                                          ▼
                        [Postgres Adapter] ─implements─▶

Vérifier :

  • Pas de cycle (A → B → A)
  • DIP respecté (domain ne dépend pas de l'infra)
  • Pas plus de 3 niveaux d'appel entre entrée et côté métier

Output : diagramme (Mermaid, ASCII, ou texte structuré).

4. Lister les trade-offs

Pour chaque décision non évidente, documenter pourquoi ce choix :

DécisionAlternative envisagéePourquoi ce choix
Async via queueSync HTTPLatence >500ms, pas critique immédiat
CQRSCRUD simpleRatio 20:1 lecture/écriture
Event sourcingStore étatBesoin audit légal

Output : ADR léger (3-5 bullets) — voir .claude/rules/10-documentation.md pour ADR formel si décision majeure.

5. Définir les tests d'architecture

Avant TDD sur le comportement, définir les tests d'architecture :

  • Tests de dépendances (ArchUnit, Dephpend, deptrac)
  • Tests de contrats (OpenAPI validation, contract tests)
  • Tests de performance cibles (latence p99, throughput)
  • Tests d'isolation multitenant si applicable

Output : liste de tests d'architecture à écrire en Phase TDD.

Livrables

À la fin de la phase Architect, produire :

  1. Diagramme dépendances (1 page max)
  2. Contrats des frontières (signatures + DTO)
  3. ADR court trade-offs principaux
  4. Liste de tests d'architecture
  5. Découpage en tâches atomiques (voir skill atomic-tasks)

Ce n'est qu'après qu'on rentre dans le TDD (Red/Green/Refactor).

Anti-patterns

Anti-patternSolution
Coder directement sans diagrammeDessiner AVANT, même 5 min suffisent
Architect phase > 2h sur feature simpleYAGNI — couper court
Contrats flous ("objet User")Types stricts, tous les champs
Ignorer les trade-offsADR court obligatoire
Architecture en vase closReview par un pair ou subagent @tech-lead

Intégration Claude Craft

  • /workflow:design — utilise ce skill par défaut
  • Agent @tech-lead — peut produire l'output Architect
  • ADR formel : voir /common:architecture-decision
  • Rule 04 SOLID + Rule 17 CQRS + Rule 14 Multitenant — consulter selon contexte

Ressources

  • obra/superpowers
  • Rule .claude/rules/04-solid-principles.md
  • Rule .claude/rules/10-documentation.md (ADR)
  • Skill atomic-tasks

Date de dernière mise à jour : 2026-04-15 Version : 1.0.0

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.