v0.2.0

What's New

Batch Optimization

• skill-forge batch --config configs/benchmarks.yaml — run optimization across multiple Skills in one command
• New skill_forge/core/batch.py module for batch orchestration

Evaluation Workflow SKILL

• skills/skill-forge-eval/SKILL.md — a reusable SKILL that guides the full optimize → evaluate → backfill → replace workflow for any SKILL.md file
• Covers L1/L2/LLM-as-Judge evaluation, auto-backfill, and safe replacement with backup

Improved Optimization Engine

• Enhanced scoring formula: 0.3 * token_score + 0.5 * quality_score + 0.2 * structure_score
• Better agent prompts for compression strategies (merge, compress, trim, consolidate)
• Improved L1 static checks and L2 benchmark keyword matching

Documentation

• Updated README.md and README_CN.md with full project documentation
• Architecture diagrams, CLI reference, and configuration guide

Optimization Results

Skill	Original	Optimized	Compression	L2 Score
wechat-docx-export	352 lines	226 lines	35.8%	4/4 (100%)
dev-workflow	2843 lines	1072 lines	62.3%	10/10 (100%)

Full Changelog: https://github.com/zccyman/skill-forge/commits/v0.2.0