Search Skill

Advanced vault search with powerful filtering using Flywheel MCP.

Purpose

Comprehensive search interface to your vault with support for:

• Folder filtering (search within specific folders)
• Tag filtering (has_tag, has_all_tags, has_any_tag)
• Frontmatter filtering (where: {key: value})
• Title search (title_contains)
• Sorting (modified, created, title)
• Pagination (limit, order)

Goes far beyond basic text search - this is semantic, metadata-aware search.

When to Use

Invoke when you want to:

• Find notes: "search for X" or "find notes about Y"
• Filter by folder: "notes in work/" or "search tech folder"
• Filter by tags: "notes with #project" or "tagged work and azure"
• Filter by metadata: "notes with status active" or "type project"
• Recent notes: "notes modified this week" or "recent changes"

Process

1. Parse User Query

Extract search parameters from natural language:

Examples:

"search for databricks"
  → title_contains: "databricks"

"notes in work/ tagged project"
  → folder: "work/", has_tag: "project"

"recent notes with status active"
  → sort_by: "modified", where: {status: "active"}

2. Call MCP Tool

Call: mcp__flywheel__search_notes
Parameters: {
  folder: "work/",
  has_tag: "project",
  title_contains: "databricks",
  sort_by: "modified",
  order: "desc",
  limit: 50
}

3. Display Results

Search Results
═══════════════════════════════════════════════

Query: databricks in work/ tagged #project
Found: 12 notes (sorted by modified, desc)

1. [[Databricks Migration Project]]
   📂 work/projects/databricks-migration.md
   📅 Modified: 2025-12-30 | Created: 2024-05-15
   🏷️ Tags: #work #project #databricks
   📋 type: project | status: active

2. [[Databricks ETL Pipeline]]
   📂 work/projects/databricks-etl.md
   📅 Modified: 2025-12-28 | Created: 2024-09-20
   🏷️ Tags: #work #project #databricks
   📋 type: project | status: completed

═══════════════════════════════════════════════

Options:
- Refine search (add/change filters)
- Show note content (preview specific note)
- Export results (CSV/markdown)

Search Parameters

Folder Filter

folder: "work/"          → Only work folder
folder: "tech/tools"     → Specific subfolder
(no folder parameter)    → All folders

Tag Filters

has_tag: "project"              → Has #project tag
has_all_tags: ["work", "azure"] → Has BOTH tags
has_any_tag: ["work", "tech"]   → Has EITHER tag

Title Filter

title_contains: "databricks"    → Title includes "databricks" (case-insensitive)
title_contains: "meeting 2024"  → Multiple words

Frontmatter Filter

where: {type: "project"}                → type == "project"
where: {status: "active", owner: "X"}   → status AND owner
where: {priority: "high"}               → priority == "high"

Sorting

sort_by: "modified"  → Most recently modified first (default)
sort_by: "created"   → Most recently created first
sort_by: "title"     → Alphabetical by title

Order

order: "desc"  → Descending (default for modified/created)
order: "asc"   → Ascending

Limit

limit: 10   → Return max 10 results
limit: 50   → Return max 50 results (default)
limit: 100  → Return max 100 results

Natural Language Examples

Simple Title Search

User: "search for databricks"
Params: {title_contains: "databricks"}
Result: All notes with "databricks" in title

Folder + Tag

User: "notes in work/ tagged project"
Params: {folder: "work/", has_tag: "project"}
Result: Project notes in work folder

Multiple Tags (AND)

User: "notes tagged work and azure"
Params: {has_all_tags: ["work", "azure"]}
Result: Notes with BOTH work AND azure tags

Recent Notes

User: "recent notes in work/"
Params: {folder: "work/", sort_by: "modified", order: "desc", limit: 20}
Result: 20 most recently modified work notes

Related Skills

• related: Find similar notes (uses tags + links)
• orphans: Find disconnected notes
• hubs: Find highly connected notes
• gaps: Find missing notes (inverse of search)

Performance

• MCP call: ~200-500ms depending on filters
• Result formatting: ~50ms for 50 results
• Total: Usually under 1 second