AgentSkillsCN

dedupe-frontmatter-links

在任何包含多行维基链接数组的 YAML frontmatter 属性中，去除重复的维基链接。例如，移除 [[ml]] 和 [[ML]] 等不区分大小写的重复链接，保留大写字母较多的版本。适用于清理 frontmatter、修复重复链接，或统一链接的大小写格式。

SKILL.md
--- frontmatter
name: dedupe-frontmatter-links
description: Deduplicate wikilinks in any YAML frontmatter property that contains multi-line wikilink arrays. Removes case-insensitive duplicates like [[ml]] and [[ML]], keeping the version with more capital letters. Use when cleaning up frontmatter, fixing duplicate links, or normalizing link casing.
allowed-tools: Read, Grep, Glob, Bash

Dedupe Frontmatter Links

Deduplicate wikilinks in YAML frontmatter properties using case-insensitive comparison.

Problem

In Obsidian, users may accidentally add duplicate wikilinks to frontmatter properties that differ only in case:

yaml
RELATED_TO:
  - "[[ml]]"
  - "[[ML]]"
  - "[[Machine Learning]]"

These duplicates clutter the frontmatter and can cause issues with queries and navigation.

Solution

This skill identifies and removes case-insensitive duplicate wikilinks, keeping the version with the most capital letters (as it's typically the more "canonical" form):

  • •Before: [[ml]], [[ML]], [[Ml]]
  • •After: [[ML]] (most capitals wins)

Instructions

Use the Python helper script at .claude/skills/dedupe-frontmatter-links/scripts/dedupe_links.py:

bash
uv run --with pyyaml .claude/skills/dedupe-frontmatter-links/scripts/dedupe_links.py <command> [args]

Helper Script Commands

bash
# Scan directory for duplicate links (dry-run, shows what would be changed)
uv run --with pyyaml .claude/skills/dedupe-frontmatter-links/scripts/dedupe_links.py scan <directory>

# Fix all duplicate links in directory
uv run --with pyyaml .claude/skills/dedupe-frontmatter-links/scripts/dedupe_links.py fix <directory>

# Check single file for duplicate links (dry-run)
uv run --with pyyaml .claude/skills/dedupe-frontmatter-links/scripts/dedupe_links.py check <file_path>

# Deduplicate links in single file
uv run --with pyyaml .claude/skills/dedupe-frontmatter-links/scripts/dedupe_links.py dedupe <file_path>

Examples

Scan SlipBox for duplicate links

bash
uv run --with pyyaml .claude/skills/dedupe-frontmatter-links/scripts/dedupe_links.py scan "02-SlipBox"

Fix all duplicates in the vault

bash
uv run --with pyyaml .claude/skills/dedupe-frontmatter-links/scripts/dedupe_links.py fix "."

Check a single note

bash
uv run --with pyyaml .claude/skills/dedupe-frontmatter-links/scripts/dedupe_links.py check "02-SlipBox/Machine Learning.md"

Deduplication Rules

  1. •Case-insensitive matching: [[ml]] and [[ML]] are considered duplicates
  2. •Prefer more capitals: When duplicates exist, keep the version with most uppercase letters
  3. •Tie-breaker: If capital count is equal, keep the first occurrence
  4. •Any wikilink array property: Processes any frontmatter property that contains multi-line wikilink arrays
  5. •Preserves order: Non-duplicate links maintain their original order

Output Format

The script outputs JSON with scan/fix results:

json
{
  "total_files_scanned": 100,
  "files_with_duplicates": 3,
  "total_duplicates_found": 7,
  "files": [
    {
      "file": "02-SlipBox/Machine Learning.md",
      "properties": {
        "RELATED_TO": {
          "before": ["[[ml]]", "[[ML]]", "[[deep learning]]"],
          "after": ["[[ML]]", "[[deep learning]]"],
          "removed": ["[[ml]]"]
        }
      }
    }
  ]
}

Notes

  • •The script automatically detects any property with multi-line wikilink array values
  • •The script skips hidden/system directories: .claude, .obsidian, .trash, .git, node_modules
  • •Always run scan first to preview changes before using fix
  • •Links with different display text (e.g., [[ML|machine learning]]) are compared by their target only
  • •Section links (e.g., [[Note#Section]]) compare only the note name portion