clarity-linter

L1 Improvement

• Rewrote the clarity linter as an English-first SOP with Prompt Architect-style constraint surfacing.
• Added structure-first guardrails, adversarial validation hooks, and confidence ceilings per Skill Forge.
• Consolidated input/output contracts into a single execution path for faster dogfooding.

---

LIBRARY-FIRST PROTOCOL (MANDATORY)

Before writing ANY code, you MUST check:

Step 1: Library Catalog

• Location: .claude/library/catalog.json
• If match >70%: REUSE or ADAPT

Step 2: Patterns Guide

• Location: .claude/docs/inventories/LIBRARY-PATTERNS-GUIDE.md
• If pattern exists: FOLLOW documented approach

Step 3: Existing Projects

• Location: D:\Projects\*
• If found: EXTRACT and adapt

Decision Matrix

Match	Action
Library >90%	REUSE directly
Library 70-90%	ADAPT minimally
Pattern exists	FOLLOW pattern
In project	EXTRACT
No match	BUILD (add to library after)

---

STANDARD OPERATING PROCEDURE

Purpose

Assess readability and cognitive load, then deliver fixes with rubric-backed evidence.

Trigger Conditions

• Positive: "clarity audit", "reduce cognitive load", "naming/readability review", "thin helper detection".
• Negative: coupling-only analysis (use connascence-analyzer), security/performance scans, or quick lint (use quick-quality-check).

Guardrails

• Structure-first package: SKILL.md, README.md, examples/, tests/, references/ kept current.
• Use explicit clarity rubric (size, indirection, call depth, duplication, explanation quality) and cite evidence.
• Apply confidence ceilings; do not overclaim automated fixes without human review when confidence <0.80.
• Store runs in memory with WHO/WHY/WHEN/PROJECT tags for traceability.

Execution Phases

1. Intent & Scope – Confirm language, repo area, and goal (audit vs fix). Load domain expertise if available.
2. Metrics Collection – Gather size, nesting, call depth, duplication, naming signals, and comment density.
3. Rubric Evaluation – Score five dimensions; classify verdict (ACCEPT ≥0.80, REFINE 0.60–0.79, REJECT <0.60).
4. Fix Planning – Rank violations by impact; propose minimal-change patches.
5. Fix Generation – Apply or draft patches; keep diffs small and reversible.
6. Validation – Rerun metrics; ensure clarity score improves and tests still pass.
7. Delivery – Summarize findings, decisions, diffs, and confidence ceiling.

Input Contract (minimum)

• target: file|directory with path.
• policy: strict|standard|lenient (defaults to standard thresholds).
• Flags: auto_fix (default false), report_format (md|json), min_score_threshold (default 0.60).

Output Format

• Metrics snapshot, rubric scores, verdict, and top violations with evidence.
• Fix plan and applied/queued patches.
• Risks, follow-ups, and memory keys used.
• Confidence: X.XX (ceiling: TYPE Y.YY).

Validation Checklist

• Trigger confirmed; correct domain expertise loaded.
• Rubric applied; evidence attached for each violation.
• If auto-fix enabled, regressions and tests rerun.
• Memory tagged and report stored.
• Confidence ceiling declared.

Integration

• Memory MCP: skills/tooling/clarity-linter/{project}/{timestamp} for reports and diffs.
• Hooks: keep evaluation under post_hook_max_ms:1000; fail fast if metrics worsen.

Confidence: 0.70 (ceiling: inference 0.70) – SOP aligned to Prompt Architect constraint extraction and Skill Forge structure-first delivery.