Architecture fitness functions

Skill Xopoko/plug-n-skills/plugins/architecture-intelligence/skills/architecture-fitness-functions

Ready-to-install skills and plugins for Codex, Claude Code, and AI coding agents: practical workflows for app delivery, architecture, research, design, and agent tooling.

Install

npx -y skills add Xopoko/plug-n-skills --skill architecture-fitness-functions

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

8 stars8 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.

What its author says it does

Copied from the file, not written here

Use when architecture choices need executable guardrails: dependency rules, cycle checks, boundary tests, ownership review gates, ADR compliance, runtime/resilience checks, and CI fitness functions.

SKILL.md

3.2 KB, as published. Nobody here has run it

Architecture Fitness Functions

Bundled commands use $PLUGIN_ROOT ($env:PLUGIN_ROOT in PowerShell; same path suffix) for the plugin root. Set it once: use the host's plugin-root variable when defined (Claude Code: PLUGIN_ROOT="$CLAUDE_PLUGIN_ROOT"), otherwise the absolute path of this plugin's root directory.

Use when architecture quality should become observable, testable, or CI-enforced.

Rule Shape

Each rule needs: intent, scope, owner, signal, mechanism, pass condition, cadence, failure action, exception path, related ADR/principle, protected quality attribute, and tradeoff.

Cadence: local, pre-commit, ci, release, scheduled, production.

Rollout: measure -> warn -> enforce for legacy systems.

Common Guardrails

Dependency direction: domain, UI, adapters, shared packages, and apps import only allowed layers.
Cycle prevention: package, namespace, module, service, or workspace cycles.
Boundary API: imports only through public entry points.
Data ownership: one bounded context owns writes for a table/topic/entity family.
Runtime coupling: sync calls, queues, cache, external APIs follow approved directions.
Observability/resilience: critical flows expose logs/metrics/traces/health checks and timeout/retry/fallback/idempotency policies.
Deployment safety: migration order, feature flags, rollback readiness, compatibility windows.
Documentation drift: architecture-changing code links ADRs, diagrams, ownership docs.
Ownership topology: architecture-significant paths have owner coverage; cross-owned changes request both owner paths or documented waiver.
Change amplification: co-change hotspots or touched-file width stay below agreed thresholds.
Conformance/metrics: forbidden edges stay absent, required edges/docs exist, cycles stay absent, high fan-out requires review.

Tooling Preference

Prefer existing project tools; add dependencies only when the boundary is important and cost is explicit.

Examples: ArchUnit/jQAssistant, NetArchTest/NDepend, dependency-cruiser/eslint/Nx graph, import-linter/pytest checks, go list, cargo metadata/clippy, SwiftPM target graph/XCTest.

Lightweight start:

python3 "$PLUGIN_ROOT/scripts/architecture_probe.py" <repo-path> --json
python3 "$PLUGIN_ROOT/scripts/architecture_probe.py" <repo-path> --json --policy <policy.json>

The probe covers top-level coupling, instability, cycles, runtime signals, ownership topology, and simple policy checks.

Workflow

State architecture principle in one sentence.
Pick the smallest measurable signal.
Choose cheapest existing mechanism.
Name protected quality attribute and tradeoff.
Define pass/fail threshold and exception policy.
Tie to ADR/rule and rollout mode.

Output

Use architecture_intelligence.fitness_plan.v1 or a compact table: Rule, Signal, Tool, Pass condition, Cadence, Owner.

Do not create brittle gates that block delivery without migration path.