Architecture advisor

Skill haytamAroui/AZ-CLAUDE/templates/skills/architecture-advisor

Evidence-based architecture and pattern decisions. Use when choosing between architectures, frameworks, patterns, databases, rendering strategies, testing approaches, or deployment targets. Triggers on: architecture, which pattern, monolith vs microservices, SSR vs SSG vs SPA, ORM vs raw SQL, REST vs GraphQL, SQL vs NoSQL, TDD vs test-after, modular monolith, serverless, edge computing, horizontal vs vertical scaling, caching strategy, state management, when to use, which is better, trade-offs, decision, scale, performance, project size, small project, large project, enterprise, startup, MVP, production-ready, best practice. Also triggers when /dream or /blueprint needs to decide stack, /copilot faces an architecture milestone, or /debate needs evidence for a technical decision. Even if the user doesn't say "architecture", use this when the task involves choosing between competing approaches for a project of a specific size or domain.From its SKILL.md

Install

npx -y skills add haytamAroui/AZ-CLAUDE --skill architecture-advisor

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.2 KB, as published. Nobody here has run it

Architecture Advisor

<instructions> Claude already knows every framework, pattern, and language. This skill doesn't teach HOW to code — it guides WHEN to use WHICH approach based on project context and evidence.

When This Fires

Any decision where the right answer depends on project scale, team size, domain, or deployment target — not just technical preference.

Step 0: Constitution Check

[ -f .claude/constitution.md ] && echo "constitution=found" || echo "no constitution"

If found: read ## Architectural Commitments before advising. Recommendations must not contradict an existing architectural commitment. If the user's question asks to deviate from a committed decision:

State the conflict explicitly: "constitution.md commits to {X}, this recommendation would change that"
Offer two paths: (a) recommendation within the constraint, (b) what amending the constitution would require
Do NOT silently recommend the deviation without flagging the conflict.

Step 1: Detect Project Context

Read these signals (skip what doesn't exist):

# Project scale
find . -name "*.ts" -o -name "*.py" -o -name "*.js" -o -name "*.go" -o -name "*.rs" 2>/dev/null | wc -l
# Framework signals
ls package.json pyproject.toml Cargo.toml go.mod pom.xml build.gradle 2>/dev/null
# Team signals
git shortlog -sn --no-merges 2>/dev/null | wc -l
# Age signals
git log --reverse --format="%ai" 2>/dev/null | head -1

Read CLAUDE.md for: Domain, Stack, Scale fields.

Classify:

SMALL: < 50 files, 1-2 contributors, < 3 months old, MVP/prototype
MEDIUM: 50-500 files, 3-10 contributors, 3-12 months, growing product
LARGE: 500+ files, 10+ contributors, 1+ year, production system

Step 2: Apply Decision Framework

For the detected scale, apply the matching evidence from references/decision-matrices.md.

Step 3: Record Decision

Append to .claude/memory/decisions.md:

## {Decision} — {date}
**Context**: {scale}, {domain}
**Choice**: {what was chosen}
**Why**: {evidence-based reason}
**Reconsider when**: {threshold}

Rules

Always state the project scale before recommending
Always cite the threshold where the recommendation changes
Always warn against the common over-engineering mistake at this scale
Never recommend microservices for projects under 50K lines or 10 contributors
Never recommend "just use Redux" without checking if the project needs global state
Never recommend TDD for throwaway prototypes — recommend test-after instead
Recommendations must be evidence-based — "X is popular" is not evidence

References

For complete decision matrices: references/decision-matrices.md For rendering strategy guide: references/rendering-decisions.md For database selection guide: references/database-decisions.md </instructions>

<output_format> Output format:

## Decision: {question}

Context: {SMALL|MEDIUM|LARGE} project, {domain}, {N} files, {N} contributors

### Recommendation: {choice}
Evidence: {why this is right for THIS context}

### When to reconsider
{thresholds that would change the answer}

### Anti-pattern warning
{what NOT to do at this scale}

</output_format>