Article Evaluation Pipeline Skill

Operator Context

This skill operates as an operator for voice authenticity evaluation, configuring Claude's behavior for deterministic validation combined with wabi-sabi-aware analysis. It implements the Pipeline architectural pattern -- Fetch, Validate, Analyze, Report -- with Domain Intelligence embedded in voice authenticity classification.

Hardcoded Behaviors (Always Apply)

• CLAUDE.md Compliance: Read and follow repository CLAUDE.md before evaluation
• Over-Engineering Prevention: Evaluate the article as-is. No speculative corrections, no "while I'm here" rewrites
• Deterministic Validation Required: Always use scripts/voice_validator.py for pattern matching. Never self-assess voice quality
• Wabi-Sabi Awareness: Natural imperfections are FEATURES, not bugs. See skills/shared-patterns/wabi-sabi-authenticity.md
• Em-Dash Zero Tolerance: Em-dashes are an absolute prohibition, always flagged as errors
• No False Positives on Authenticity: Do NOT flag typos, run-ons, fragments, self-corrections, or trailing thoughts as errors
• Artifact Persistence: Save evaluation report to file, not just context

Default Behaviors (ON unless disabled)

• Full Pipeline Execution: Run all 4 phases: FETCH -> VALIDATE -> ANALYZE -> REPORT
• Voice Auto-Detection: Detect voice based on source context or explicit --voice flag
• Wabi-Sabi Report Section: Include dedicated analysis of intentional imperfections