name: readme-sync
description: Automatically synchronize README.md with actual project components. Scans .claude directory and updates Agents, Skills, Hooks, Commands sections.

README Sync Skill

Automatically keeps README.md synchronized with actual project components. Scans the .claude directory structure and updates documentation to match reality.

Usage

/readme-sync [--section <name>] [--validate] [--dry-run]

Options

What Gets Synced

1. Agents Section

Scans .claude/agents/*.md and updates:
- Agent count
- Agent table with names, purposes, keywords
- Auto-detects from frontmatter: name, description

2. Skills Section

Scans .claude/skills/*/SKILL.md and .claude/skills/*.md:
- Skill count
- Skill table with usage, description
- Auto-detects from frontmatter

3. Hooks Section

Scans .claude/hooks/*.sh and .claude/hooks/*.md:
- Hook count
- Hook table with events, purposes
- Auto-detects from frontmatter or comments

4. Commands Section

Scans .claude/commands/*.md and .claude/commands/*/:
- Command count
- Command table with descriptions
- Includes subcommand files

5. Stats Section

Updates summary statistics:

| Category | Count |
|----------|-------|
| Agents | 16 |
| Skills | 14 |
| Hooks | 4 |
| Commands | 2 |
| Templates | 5 |
| **Total** | **41** |

Workflow

Step 1: Scan Components

# Scan all .claude directories
find .claude/agents -name "*.md" -type f
find .claude/skills -name "SKILL.md" -o -name "*.md" -type f
find .claude/hooks -name "*.sh" -o -name "*.md" -type f
find .claude/commands -name "*.md" -type f

Step 2: Extract Metadata

For each file, extract:

# From frontmatter
name: component-name
description: Brief description
tools: [Tool1, Tool2]  # for agents
keywords: [key1, key2]

Step 3: Generate Section Content

## Sub-Agents (16)

### Development & Analysis

| Agent | Purpose | Keywords |
|-------|---------|----------|
| `project-analyzer` | Analyze project structure | "analyze", "structure" |
| `code-reviewer` | Review code quality | "review", "PR" |
...

Step 4: Update README

• Find existing section by header
• Replace content between headers
• Preserve custom content outside synced sections

Section Markers

Use these markers to define sync boundaries:

<!-- README-SYNC:agents:start -->
## Sub-Agents (16)
...content auto-generated...
<!-- README-SYNC:agents:end -->

<!-- README-SYNC:skills:start -->
## Skills (14)
...content auto-generated...
<!-- README-SYNC:skills:end -->

Without markers, the skill will find sections by ## Sub-Agents or ## Agents headers.

Output Examples

Sync Report

📝 README-SYNC: Synchronizing README.md...

[1/4] Scanning agents...
      Found: 16 agents
      New: file-explorer (+1)
      Removed: none

[2/4] Scanning skills...
      Found: 14 skills
      New: agile-sync, readme-sync (+2)
      Removed: none

[3/4] Scanning hooks...
      Found: 4 hooks
      New: auto-doc-sync (+1)
      Removed: none

[4/4] Updating stats...
      Total: 36 → 40 (+4)

📝 README-SYNC: Complete!
   Sections updated: 4
   Components added: 4
   Components removed: 0

Validation Report

📝 README-SYNC: Validation Report

## Agents
✅ 16 agents in README matches 16 files
⚠️ `old-agent` in README but file not found

## Skills
❌ README shows 12 skills, found 14 files
   Missing in README:
   - agile-sync
   - readme-sync

## Hooks
✅ All hooks accounted for

## Stats
❌ Stats section outdated
   README: 33 total
   Actual: 40 total

Issues: 3
Recommendations:
- Run `/readme-sync` to fix mismatches
- Remove reference to `old-agent`

Configuration

Auto-Sync Settings

{
  "readme-sync": {
    "auto_on_change": true,
    "sections": ["agents", "skills", "hooks", "commands", "stats"],
    "preserve_custom": true,
    "use_markers": true,
    "table_format": "markdown"
  }
}

Section Templates

# .claude/readme-sync-config.yml
sections:
  agents:
    header: "## Sub-Agents ({count})"
    columns:
      - name: "Agent"
        source: "name"
        format: "`{value}`"
      - name: "Purpose"
        source: "description"
        truncate: 50
      - name: "Keywords"
        source: "keywords"
        format: "\"{value}\""

  skills:
    header: "## Skills ({count})"
    group_by: "category"
    columns:
      - name: "Skill"
        source: "name"
      - name: "Usage"
        source: "usage"
      - name: "Description"
        source: "description"

Integration

With auto-doc-sync Hook

# Hook triggers readme-sync when .claude files change
# In hooks/auto-doc-sync.sh:
if [[ "$TOOL_INPUT" == *".claude/"* ]]; then
    # Trigger README sync suggestion
fi

With /agile-sync

# agile-sync includes readme-sync as step 3
/agile-sync  # Includes README stats update

Manual Trigger

# Full sync
/readme-sync

# Specific section
/readme-sync --section skills

# Validate only
/readme-sync --validate

Best Practices

DO

• ✅ Use section markers for precise updates
• ✅ Run after adding new agents/skills
• ✅ Include in PR checklist
• ✅ Keep frontmatter accurate in source files

DON'T

• ❌ Manually edit synced sections (will be overwritten)
• ❌ Remove section markers
• ❌ Skip validation before releases

Troubleshooting

"Section not found in README"

# Add section markers or headers:
## Sub-Agents
<!-- or -->
<!-- README-SYNC:agents:start -->
<!-- README-SYNC:agents:end -->

"Component not detected"

# Ensure frontmatter is valid:
---
name: my-component
description: Brief description
---

"Stats mismatch"

# Force full recalculation:
/readme-sync --section stats

Related Skills