auto-updater（自动更新程序）

Auto Updater

Overview

auto-updater automatically applies improvements to skills and ecosystem components based on identified patterns and learnings.

Purpose: Automated application of validated improvements across ecosystem

The 5-Step Auto-Update Workflow:

1. Identify Improvements - Gather recommendations from reviews and learnings
2. Assess Safety - Determine which can be safely automated
3. Apply Updates - Implement improvements automatically
4. Validate Changes - Ensure improvements effective, no regressions
5. Rollback if Needed - Revert changes if validation fails

Safety: Always validates before finalizing, can rollback

When to Use

• Applying systematic improvements across multiple skills
• Implementing guideline updates ecosystem-wide
• Automating common enhancement patterns
• Bulk updates (e.g., add Quick Reference to all skills missing it)

Auto-Update Workflow

Step 1: Identify Improvements

Sources:

• system-reviewer recommendations
• best-practices-learner documented patterns
• review-multi common findings
• Manual improvement requests

Output: List of potential improvements

Time: 15-30 minutes

Step 2: Assess Safety

Safe to Automate:

• Structural additions (add Quick Reference section)
• Content additions (add examples in standard locations)
• Format standardization (consistent heading levels)
• Documentation updates (README enhancements)

NOT Safe to Automate:

• Logic changes (requires understanding context)
• Content rewrites (needs judgment)
• Major refactoring (risk too high)
• Custom implementations

Output: Classified improvements (auto-safe vs manual-only)

Time: 20-40 minutes

Step 3: Apply Updates

Process:

1. Backup affected skills (git commit or copy)
2. Apply improvement to each skill
3. Log changes made
4. Track success/failure per skill

Approach: One skill at a time, validate each before moving to next

Time: Varies by improvement and skill count

Step 4: Validate Changes

For Each Updated Skill:

1. Run skill-validator (pass/fail)
2. Run review-multi structure check (score maintained?)
3. Visual inspection (looks correct?)
4. Mark as validated or flagged for review

Output: Validation results per skill

Time: 10-15 minutes per skill

Step 5: Rollback if Needed

If Validation Fails:

1. Identify which skill failed
2. Restore from backup (git revert or copy back)
3. Analyze why it failed
4. Mark improvement as manual-only for that skill

Output: Rolled back skill, failure analysis

Example Auto-Update

Auto-Update: Add Quick Reference to All Skills Missing It

Step 1: Identify
- Improvements: Add Quick Reference section
- Target Skills: planning-architect, task-development, todo-management
- Count: 3 skills to update

Step 2: Assess Safety
- ✅ Safe: Adding new section (doesn't modify existing content)
- ✅ Safe: Standard format (use template)
- ✅ Safe: Low risk (can validate easily)
- Decision: Auto-update approved

Step 3: Apply
- Backup: Git commit all 3 skills
- Apply to planning-architect: ✅ Success
- Apply to task-development: ✅ Success
- Apply to todo-management: ✅ Success
- Changes: 3/3 skills updated

Step 4: Validate
- planning-architect: 5/5 structure (maintained)
- task-development: 5/5 structure (maintained)
- todo-management: 5/5 structure (maintained)
- All validations: ✅ PASS

Step 5: Rollback
- Not needed (all validations passed)

Result: ✅ 3 skills successfully auto-updated
Time: 90 minutes (vs 3-4 hours manual)
Impact: 100% Quick Reference coverage achieved
Quality: All skills maintained 5/5 scores

Quick Reference

5-Step Auto-Update Workflow

Safe vs Unsafe Automation

Safe to Automate:

• Adding standard sections
• Format standardization
• Documentation additions
• Structural improvements (following patterns)

NOT Safe:

• Logic changes
• Content rewrites
• Major refactoring
• Custom implementations

Rule: If requires judgment or understanding → Manual only

auto-updater enables safe, validated, automated improvement application across multiple skills simultaneously.