Notre avis
Analyse constructive en tant qu'avocat du diable sur des décisions techniques, choix d'architecture et compromis, en identifiant les risques et alternatives.
Points forts
- Structure rigoureuse en niveaux (critique décision, portée, test premier jour).
- Encourage la recherche proactive de lacunes et de risques cachés.
- Ton respectueux et factuel, évite l'inflation artificielle des risques.
- Inclut une évaluation de la réversibilité des décisions.
Limites
- Nécessite un document ou une décision existante pour fonctionner.
- Peut être perçu comme excessivement critique si non appliqué avec nuance.
- Ne remplace pas une analyse de risque approfondie par un expert humain.
Utilisez cette compétence lorsque vous évaluez un choix technologique, rédigez une ADR, ou souhaitez stress-tester une décision avant de l'engager dans la documentation.
Évitez de l'utiliser pour des décisions triviales ou lorsque la vitesse est prioritaire sur l'exhaustivité.
Analyse de sécurité
SûrThis skill is purely analytical, providing a structured critique framework for decisions; it does not invoke any tools or execute code, posing no security risk.
Aucun point d'attention détecté
Exemples
We're considering PostgreSQL for our event-sourcing system. Perform a devil's advocate analysis on this choice, considering scalability, complexity, and operational cost.Here is my draft ADR for adopting microservices. Act as devil's advocate and identify gaps, risks, and hidden assumptions.We're prioritizing user authentication overhaul over performance improvements. Challenge this decision as devil's advocate.name: devil-advocate description: > Performs constructive devil's advocate analysis on decisions, technology choices, and architectural approaches. Use when evaluating tradeoffs, challenging assumptions, or when a decision needs stress-testing before committing to it in documentation.
Devil's Advocate Analysis (Enhanced)
When performing devil's advocate analysis, follow this structured approach:
Trigger Conditions
Activate this skill when:
- A technology choice is being made (database, framework, hosting, etc.)
- An architectural decision is being documented in an ADR
- A build-vs-buy decision arises
- The user is confident in an approach that has non-obvious risks
- Feature prioritization involves significant tradeoffs
- After every document generation (automatic — scope critique)
Analysis Structure
Level 1: Decision Critique
1. Steel-man the current position
Before attacking, present the BEST case for the current choice. This shows fairness and ensures the critique is against a strong version of the argument.
2. Identify the three biggest risks
For each risk:
- Probability: How likely is this to happen? (cite evidence)
- Impact: What happens if it does?
- Detectability: Will we know before it's too late?
- Mitigation: Can we reduce the risk while keeping the choice?
3. Name the best alternative
Not a strawman — the genuine best alternative approach. Explain when it would be the right choice.
4. Identify reversibility
- Easily reversible: Low-cost to change later → proceed with less analysis
- Costly to reverse: Significant migration → more analysis warranted
- Irreversible: Permanent commitment → maximum scrutiny required
5. Deliver verdict
Always end with a clear, actionable recommendation. Never leave the user in analysis paralysis.
Level 2: Scope Critique (Run after every document)
Ask these SPECIFIC questions:
- "What did we FORGET?" — actively search for missing requirements
- "What breaks when this fails?" — failure mode analysis
- "What does the user need that they didn't ask for?" — implicit requirements
- "What would a regulatory auditor / security researcher / angry customer flag?"
Level 3: "First Day on the Job" Test
Imagine a developer reading this doc pack on Day 1:
- Can they set up the project without asking questions?
- Do they know which doc to read for which task?
- Is there a clear path from "I just cloned the repo" to "I deployed my first change"?
- Are gotchas and non-obvious constraints highlighted?
Minimum Gaps Rule
The devil's advocate MUST identify at least 2 potential gaps or concerns per document. If you genuinely can't find any, state: "No gaps identified — this document is comprehensive." This forces active gap-seeking rather than rubber-stamping.
Tone
- Respectful but direct
- Evidence-based, not opinion-based
- Acknowledge when the current choice IS the best option
- Never artificially inflate risks to seem thorough
- Focus on SCOPE gaps (what's missing?) not just DECISION quality (is this the right choice?)
Expert Next.js App Router
Developpement
Un skill qui transforme Claude en expert Next.js App Router.
Générateur de README
Developpement
Crée des README.md professionnels et complets pour vos projets.
Rédacteur de Documentation API
Developpement
Génère de la documentation API complète au format OpenAPI/Swagger.