architecture-review

Assesses architecture decisions, ADR compliance, and coupling

Night Market Skill — ported from claude-night-market/pensive. For the full experience with agents, hooks, and commands, install the Claude Code plugin.

Table of Contents

• Quick Start
• When to Use
• Progressive Loading
• Required TodoWrite Items
• Workflow
• Step 1: Establish Context (arch-review:context-established)
• Step 2: ADR Audit (arch-review:adr-audit)
• Step 3: Interaction Mapping (arch-review:interaction-mapping)
• Step 4: Principle Checks (arch-review:principle-checks)
• Step 5: Risks and Actions (arch-review:risks-actions)
• Testing

Testing

Run pytest plugins/pensive/tests/skills/test_architecture_review.py to verify review logic.

• Architecture Principles Checklist
• Coupling
• Cohesion
• Layering
• Evolution

Architecture Review Workflow

Architecture assessment against ADRs and design principles.

Quick Start

bash

/architecture-review

When To Use

• Approving reimplementations.
• Large-scale refactoring reviews.
• System design changes.
• New module/service introduction.
• Dependency restructuring.

When NOT To Use

• Selecting architecture paradigms - use archetypes skills
• API surface review - use api-review
• Selecting architecture paradigms - use archetypes skills
• API surface review - use api-review

Progressive Loading

Load modules based on review scope:

• modules/adr-audit.md (~400 tokens): ADR verification and documentation.
• modules/coupling-analysis.md (~450 tokens): Dependency analysis and boundary violations.
• modules/principle-checks.md (~500 tokens): Code quality, security, and performance.
• modules/fpf-methodology.md (~800 tokens): FPF (Functional, Practical, Foundation) multi-perspective review methodology.

Load all modules for full reviews. For focused reviews, load only relevant modules.

Required TodoWrite Items

1. arch-review:context-established: Repository, branch, motivation.
2. arch-review:adr-audit: ADR verification and new ADR needs.
3. arch-review:interaction-mapping: Module coupling analysis.
4. arch-review:invariant-check: Invariant conflict detection and 3-option analysis.
5. arch-review:principle-checks: LoD, security, performance.
6. arch-review:risks-actions: Recommendation and follow-ups.

Workflow

Step 1: Establish Context (arch-review:context-established)

Confirm repository and branch:

bash

pwd
git status -sb

Document:

• Feature/bug/epic motivating review.
• Affected subsystems.
• Architectural intent from README/docs.
• Design trade-off assumptions.

Step 2: ADR Audit (arch-review:adr-audit)

Load: modules/adr-audit.md

• Locate ADRs in project.
• Verify required sections.
• Check status flow.
• Confirm immutability compliance.
• Flag need for new ADRs.

Step 3: Interaction Mapping (arch-review:interaction-mapping)

Load: modules/coupling-analysis.md

• Diagram before/after module interactions.
• Verify composition boundaries.
• Check data ownership clarity.
• Validate dependency flow direction.
• Identify coupling violations.

Step 3.5: Invariant Conflict Detection (arch-review:invariant-check)

Before checking principles, identify whether the changes conflict with existing design invariants. This is the highest-judgment step in architecture review — models get this wrong more often than any other call.

Identify existing invariants:

1. Scan ADRs for recorded decisions still in "accepted" status
2. Check module boundaries (are imports crossing layers that previously didn't?)
3. Check data flow direction (does data now flow in a new direction?)
4. Check API contracts (are public interfaces changing shape?)
5. Check structural patterns (is a new pattern being introduced alongside an existing one?)

bash

# Detect boundary crossings in changed files
git diff --name-only | while read f; do
  head -20 "$f" 2>/dev/null | rg "^(import|from|use |require)" || true
done

When a conflict is detected:

Do NOT recommend a resolution. Present the three options and escalate to human judgment:

Option	When Right	When Wrong
Preserve invariant (reject feature)	Invariant simplifies many things; feature is marginal	Feature is genuinely needed and invariant is stale
Layer on top (add inelegantly)	Feature is needed; invariant still valuable; imperfection is OK	Layering creates a maintenance trap that will compound
Revise invariant (change the design)	Genuine new learning invalidates the original reasoning	You're "cleaning up" a decision you don't fully understand

Output format:

markdown

### Invariant Conflicts

[I1] **[Invariant name]** — [what decision it represents]
- **Conflict**: [what change clashes]
- **Options**: Preserve / Layer / Revise
- **Recommendation**: ESCALATE TO HUMAN
- **Risk if wrong**: [what compounds]

Why this matters: Bad invariant decisions compound. After a few wrong calls the codebase becomes unsalvageable. This is a judgment problem, not a context problem — the agent should surface it, not solve it.

Step 4: Principle Checks (arch-review:principle-checks)

Load: modules/principle-checks.md

• Law of Demeter.
• Anti-slop patterns.
• Security (input validation, least privilege).
• Performance (N+1 queries, caching).

Step 5: Risks and Actions (arch-review:risks-actions)

Summarize using imbue:diff-analysis/modules/risk-assessment-framework:

• Current vs proposed architecture.
• Business impact.
• Technical debt implications.

List follow-ups with owners and dates.

Provide recommendation:

• Approve: Architecture sound.
• Approve with actions: Minor issues to address.
• Block: Fundamental problems requiring redesign.

Architecture Principles Checklist

Coupling

• Dependencies follow defined boundaries.
• No circular dependencies.
• Extension points used properly.
• Abstractions don't leak.

Cohesion

• Related functionality grouped.
• Single responsibility per module.
• Clear module purposes.

Layering

• Layers have clear responsibilities.
• Dependencies flow downward.
• No layer bypassing.

Invariants

• Existing design invariants identified.
• Conflicts between changes and invariants surfaced.
• Three-option analysis (preserve/layer/revise) presented.
• Invariant changes escalated to human judgment.
• No silent invariant revisions in the diff.

Evolution

• Changes are reversible.
• Migration paths are clear.
• ADRs document decisions.