documentation-update（文档更新）

Documentation Update Skill

When to Use This Skill

• Adding entries to README.md tables after integration
• Updating skill/command/agent indices
• Maintaining table of contents
• Cross-referencing new content
• Keeping documentation in sync with codebase

What This Skill Does

Provides systematic approaches for:

• Table Updates - Adding rows while preserving format
• Alphabetical Insertion - Maintaining sorted order
• Link Validation - Ensuring references are correct
• Format Preservation - Matching existing markdown style
• Duplicate Prevention - Avoiding redundant entries

Table Update Patterns

Pattern 1: Adding to Markdown Tables

Identify Table Structure:

| Column1 | Column2 | Column3 |
|---------|---------|---------|
| Value1  | Value2  | Value3  |

Steps:

1. Locate table - Search for table header
2. Extract format - Note column alignment (left/center/right)
3. Create new row - Match column count and format
4. Insert position - Alphabetical or by category
5. Validate - Ensure pipes align, no extra/missing columns

Example - Adding Skill to README:

Original:

| Skill | Purpose | Use When |
|-------|---------|----------|
| **skill-creator** | Creates new skills | Building new automation |
| **skill-orchestrator** | Coordinates multiple skills | Complex workflows |

New Entry Data:

Name: using-git-worktrees
Purpose: Isolated workspace management
Use When: Feature work needing isolation

Updated (alphabetically inserted):

| Skill | Purpose | Use When |
|-------|---------|----------|
| **skill-creator** | Creates new skills | Building new automation |
| **skill-orchestrator** | Coordinates multiple skills | Complex workflows |
| **using-git-worktrees** | Isolated workspace management | Feature work needing isolation |

Pattern 2: Alphabetical Insertion

Algorithm:

function insertAlphabetically(table, newEntry):
  rows = table.split("\n")
  headerRow = rows[0]
  separatorRow = rows[1]
  dataRows = rows[2..]

  // Extract first column values for sorting
  existingEntries = dataRows.map(row => extractFirstColumn(row))

  // Find insertion point
  insertionIndex = 0
  for i, entry in existingEntries:
    if newEntry < entry:
      insertionIndex = i
      break
    else:
      insertionIndex = i + 1

  // Insert new row
  dataRows.insert(insertionIndex, formatRow(newEntry))

  // Rebuild table
  return [headerRow, separatorRow, ...dataRows].join("\n")

Pattern 3: Category-Based Tables

For tables organized by category (not alphabetical):

Example - Command Categories:

#### Core Workflow Commands
- **start-session.md** - Initialize development session
- **close-session.md** - End session with summary

#### Quality Assurance Commands
- **test-all.md** - Execute test suites
- **lint-fixes.md** - Auto-fix style issues

Insertion Logic:

1. Determine category from content/metadata
2. Locate correct category section
3. Add to end of that category (or alphabetically within)
4. Maintain bullet/numbering format

Link Update Patterns

Pattern 4: Updating Cross-References

Find and Update Links:

Old: See [Document 07](docs/best-practices/07-Quick-Reference.md)
New: See [Document 07](docs/best-practices/07-Quick-Reference.md) and [Document 08](docs/best-practices/08-Skills-Guide.md)

Validation:

# Check link target exists
!test -f docs/best-practices/08-Skills-Guide.md && echo "✅ Valid"

Pattern 5: Creating Index Files

For skills/README.md:

# Skills Directory

## Available Skills

[Auto-generated table from skills/*/SKILL.md files]

| Skill | Description | Category |
|-------|-------------|----------|
[Rows auto-populated from frontmatter]

## Usage

[Standard usage instructions]

## Creating Skills

[Link to templates and guides]

Generation Logic:

function generateSkillsIndex():
  skills = findAll("skills/*/SKILL.md")

  table = createTable(["Skill", "Description", "Category"])

  for skill in skills:
    frontmatter = extractFrontmatter(skill)
    name = frontmatter.name
    description = frontmatter.description
    category = determineCategory(skill)

    table.addRow([name, description, category])

  table.sortAlphabetically()

  return table

Format Preservation

Matching Existing Style

Detect Style:

- Bold pattern: **text** or __text__
- Link pattern: [text](url) or [text][ref]
- List style: - item or * item or 1. item
- Code blocks: ```lang or ~~~lang
- Headers: # H1 or ## H2 (ATX style)

Apply Consistently:

# If existing entries use:
| **bold-name** | Description |

# New entry must use:
| **new-entry** | Description |

# NOT:
| new-entry | Description |

Column Alignment

Preserve alignment indicators:

| Left | Center | Right |
|:-----|:------:|------:|
| L1   |   C1   |    R1 |
| L2   |   C2   |    R2 |

New row must match:

| New  |  Data  |   123 |

Duplicate Prevention

Check Before Adding

function isDuplicate(table, newEntry):
  existingEntries = extractEntriesFromTable(table)

  for entry in existingEntries:
    if entry.name == newEntry.name:
      return true
    if entry.path == newEntry.path:
      return true

  return false

Action on Duplicate:

• Exact match: Skip, note in update report
• Similar match: Flag for review
• Same name, different path: Warning, possible conflict

Example Usage

Example 1: Add 6 Skills to README

Input:

skills = [
  {name: "content-research-writer", purpose: "Writing assistance", useWhen: "Writing articles"},
  {name: "root-cause-tracing", purpose: "Systematic debugging", useWhen: "Tracing bugs"},
  {name: "sharing-skills", purpose: "Contribute skills upstream", useWhen: "Sharing patterns"},
  {name: "subagent-driven-development", purpose: "Execute plans with subagents", useWhen: "Plan execution"},
  {name: "using-git-worktrees", purpose: "Isolated workspace management", useWhen: "Feature isolation"},
  {name: "using-superpowers", purpose: "Meta-skill for skill discovery", useWhen: "Starting conversations"}
]

Process:

1. Read README.md
2. Locate "Pre-Built Skills" table
3. For each skill:

- Check for duplicates - Format as table row - Insert alphabetically

1. Write updated README.md
2. Verify table structure intact

Output Report:

### README.md Updates
- ✅ Added 6 skills to Pre-Built Skills table
- ✅ Alphabetical order maintained
- ✅ No duplicates created
- ✅ Table format preserved
- ✅ All links validated

Example 2: Create skills/README.md Index

Input: List of all skills in skills/*/SKILL.md

Process:

1. Scan skills/ directory
2. Read each SKILL.md file
3. Extract frontmatter (name, description)
4. Categorize by type (workflow/content/meta)
5. Generate organized tables
6. Add usage instructions
7. Write to skills/README.md

Output: Comprehensive index file with categorized skill tables

Example 3: Update Command List

Input: New command integration-process.md

Process:

1. Determine category: "Integration & Maintenance Commands"
2. Extract description from frontmatter
3. Locate category section in README
4. Format as bullet item with link
5. Insert (alphabetically or at end)
6. Save README.md

Validation Checks

Pre-Update Validation

• Target file exists and is writable
• Table structure is valid
• New entry has all required fields
• No duplicate entries
• Links reference existing files

Post-Update Validation

• Table structure still valid
• No broken pipes or misaligned columns
• All links still work
• Alphabetical order maintained (if applicable)
• No extra blank lines introduced

Validation Commands

# Check table structure
!grep -A 5 "| Skill | Purpose |" README.md | head -10

# Verify link targets
!for link in $(grep -o "\](.*\.md)" README.md | tr -d ']()'); do
  test -f "$link" || echo "Broken: $link"
done

# Count table rows
!grep -c "|.*|.*|" README.md

Error Handling

Malformed Table

Error: Table has inconsistent column counts
Action: Report error, do not modify
Fix: Manually correct table structure first

File Not Writable

Error: Permission denied writing to README.md
Action: Log error, skip update
Fallback: Create pending-updates.md with changes

Duplicate Entry

Warning: Skill "using-git-worktrees" already in table
Action: Skip insertion, note in report
Info: Entry exists at line 156

Integration with Commands

Used By

• /integration-update-docs - Primary documentation updater
• /maintenance-plan-update - DEVELOPMENT_PLAN.md updates
• Custom documentation workflows

Usage Pattern

# In integration-update-docs command

After integration:
  1. Load integration report
  2. For each integrated file:
     - Use documentation-update skill
     - Update appropriate doc files
     - Validate changes
  3. Generate doc update report

Best Practices

DO

• ✅ Always backup before modifying
• ✅ Validate table structure before and after
• ✅ Preserve existing formatting
• ✅ Check for duplicates
• ✅ Verify all links
• ✅ Maintain alphabetical order
• ✅ Test with git diff before committing

DON'T

• ❌ Assume table structure
• ❌ Mix formatting styles
• ❌ Skip duplicate checks
• ❌ Modify without validation
• ❌ Break existing links
• ❌ Introduce trailing whitespace

Testing Recommendations

Test with:

1. Simple table - Add single entry to clean table
2. Complex table - Multi-column with formatting
3. Edge cases - Empty table, single row, malformed
4. Bulk updates - Add 10+ entries at once
5. Recovery - Test rollback on error

Expected behavior:

• Clean tables: 100% success
• Malformed tables: Detect and report, no corruption
• Duplicates: Detect and skip appropriately

Version History

1.0 (2025-11-23)

• Initial documentation update skill
• Table manipulation logic
• Link validation
• Format preservation
• Duplicate prevention

Skill Status: Production Ready Integration: Used by /integration-update-docs Dependencies: None (standalone logic)