skill-packager

原文涵盖的主题

• Skill Packager Skill
• Purpose
• Grounding Checkpoint (Archetype 1 Mitigation)
• Uncertainty Escalation (Archetype 2 Mitigation)
• Context Scope (Archetype 3 Mitigation)
• Workflow Steps
• Step 1: Validate Skill Structure (Grounding)
• Step 2: Security Check

Single responsibility: Package completed skill directories into ZIP files ready for upload to Claude AI. (BP-4) Before executing, VERIFY: [ ] Skill directory exists with required structure [ ] SKILL.md is present and non-empty [ ] At least one reference file exists [ ] No sensitive data in skill directory [ ] Output path for ZIP is writable DO NOT package without validating skill structure. ASK USER instead of guessing when: Skill structure incomplete - proceed anyway? Large files detected - include or exclude? Sensitive patterns found (API keys, passwords) Multiple skill directories - which to package? NEVER package potentially sensitive content without review. | Context Type | Included | Excluded | |--------------|----------|----------| | RELEVANT | Skill directory contents, package config | Other skills | | PERIPHERAL | Package size estimates | Source data | | DISTRACTOR | Build process details | Scraping history | Option A: With skill-seekers Option B: Manual packaging On error: 1. PAUSE - Don't overwrite existing ZIP 2. DIAGNOSE - Check error type: Missing files → Run skill-builder first ZIP error → Check disk space, permissions Large size → Exclude unnecessary files Sensitive

这里只展示节选，请以完整原文和当前要求为准。