Claude Code Plugins

Community-maintained marketplace

Feedback

connascence-quality-gate

@DNYoussef/context-cascade
8
0

Enforce connascence-based quality gates by detecting harmful coupling and demanding refactoring plans with evidence.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name connascence-quality-gate
description Enforce connascence-based quality gates by detecting harmful coupling and demanding refactoring plans with evidence.
allowed-tools Read, Write, Edit, Bash, Glob, Grep, Task, TodoWrite
model sonnet
x-version 3.2.0
x-category quality
x-vcl-compliance v3.1.1
x-cognitive-frames HON, MOR, COM, CLS, EVD, ASP, SPC

STANDARD OPERATING PROCEDURE

Purpose

Assess code for forms of connascence (name, type, meaning, timing, algorithm, position, execution) and enforce remediation gates before merge.

Trigger Conditions

  • Positive: reviews focused on coupling/maintainability risk, architectural refactors, or release gates that target hidden dependencies.
  • Negative: pure functionality debugging (use functionality-audit) or cosmetic linting only.

Guardrails

  • Confidence ceiling: include Confidence: X.XX (ceiling: TYPE Y.YY) with ceilings {inference/report 0.70, research 0.85, observation/definition 0.95}.
  • Evidence-first: Each connascence finding must cite file:line, connascence type, and impact (risk or cost) with a proposed decoupling strategy.
  • Structure-first: Maintain examples and tests that show both detection and approved remediation patterns.
  • Adversarial validation: Challenge borderline cases (intentional coupling, domain invariants) and mark waivers explicitly.

Execution Phases

  1. Context & Scope
    • Identify modules or features under review and allowed coupling (domain invariants, protocol guarantees).
    • Exclude generated/vendor code unless coupling leaks into product code.
  2. Detection & Classification
    • Scan for each connascence type; record evidence with severity (high if change ripple or hidden temporal ordering).
    • Note systemic patterns (shared globals, positional parameters, synchronized deployments).
  3. Remediation Design
    • Propose refactors (interfaces, adapters, events, typed contracts, configuration) and expected impact.
    • Sequence fixes to minimize blast radius; align with product timelines.
  4. Validation & Gate Decision
    • Require at least one mitigation path per critical finding before approval.
    • Document waivers with rationale and expiry; state confidence with ceiling.

Output Format

  • Scope summary and allowed/forbidden coupling rules.
  • Finding list with connascence type, file:line evidence, impact, and remediation.
  • Gate decision (approve/block/waiver) with reasoning.
  • Confidence statement using ceiling syntax.

Validation Checklist

  • Scope and exemptions captured.
  • Each finding includes connascence type, evidence, and remediation.
  • Waivers include expiry and owner; blockers flagged.
  • Examples/tests updated or queued.
  • Confidence ceiling provided; English-only output.

Confidence: 0.71 (ceiling: inference 0.70) - SOP rewritten with Prompt Architect confidence discipline and Skill Forge structure-first guardrails.