/note — Quick Capture

Create a SlashNote sticky note instantly from the user's input. Zero friction — one command, one tool call, done.

Usage

/note <text>

Auto-Detection Rules

Analyze the user's input using a confidence scoring system. Each signal adds weight — the category with the highest score wins. Minimum 2 points to activate a category; otherwise default to plain note.

Checklist (peach) — score 2+ to activate

Format: Convert each item to - [ ] Item. If comma-separated, split by commas. No heading needed.

Bug (pink) — score 2+ to activate

Format: # Bug: <concise title (max 8 words)> then body text. The word "fix" alone does NOT trigger bug — it's too common in normal usage.

Idea (blue) — score 2+ to activate

Format: # Idea heading + the text as body.

Code (purple) — score 2+ to activate

Format: Wrap in code block with detected language. Add # Snippet heading if content is multi-line.

Link (yellow + bookmark) — score 2+ to activate

Format: If input is URL-only → # Bookmark heading + URL as body. If URL + description → description as title + URL below. Color stays yellow but content is formatted.

Default (yellow) — fallback

When no category reaches 2 points, or input is short (< 5 words):

• Plain text, no heading, no formatting
• Color: yellow

Auto-Title

For notes longer than 20 words that don't get a heading from category detection:

• Generate a concise title (max 6 words) from the content
• Add as # <Title> heading
• This makes notes scannable in the list view

Short notes (< 20 words) → no auto-title needed.

Tie-Breaking

If multiple categories score equally:

1. Checklist > Bug > Idea > Code > Link (priority order)
2. If still ambiguous, default to yellow plain text

Behavior

1. Parse the user's input after /note
2. Score each category using signals above
3. Pick the highest-scoring category (minimum 2 points)
4. Apply auto-title if needed (long notes without heading)
5. Call mcp__slashnote__create_note with detected color and formatted content
6. Confirm to the user: "Created [color] note: [first line preview]"

Examples

Input: /note Fix race condition in WebSocket handler Scoring: Bug gets 0 ("fix" alone = 0 points) → Yellow plain text

Input: /note Bug: race condition crashes WebSocket handler Scoring: Bug gets 6 → Pink bug note Result: # Bug: Race condition crashes WebSocket handler

Input: /note - Buy milk - Review PR - Deploy staging Scoring: Checklist gets 4 → Peach checklist with 3 checkboxes

Input: /note buy milk, review PR, deploy staging, send email Scoring: Checklist gets 3 → Peach checklist with 4 checkboxes

Input: /note Use LRU cache for API responses Scoring: No category reaches 2 → Yellow plain text

Input: /note Maybe we should try SSE instead of WebSockets Scoring: Idea gets 5 → Blue idea with # Idea heading

Input: /note func validate(token: String) -> Bool Scoring: Code gets 4 → Purple Swift code block

Input: /note https://github.com/anthropics/claude-code Scoring: Link gets 3 → Yellow bookmark Result: # Bookmark heading + URL

Input: /note Great article about SwiftUI performance https://swiftui-lab.com/performance Scoring: Link gets 3 → Yellow with description as title Result: # SwiftUI Performance Article heading + URL below

Input: /note We discussed the migration strategy with the team today. The consensus was to use a phased approach starting with the auth module, then moving to the API layer, and finally the UI components. Timeline is 3 sprints. Auto-title applied (>20 words, no category) → Yellow with generated title Result: # Migration Strategy — Phased Approach heading + full text

Important

• Keep it fast — one tool call, no extra questions
• Do NOT ask the user for confirmation, color choice, or anything else
• Confidence scoring happens in your head — do not show scores to the user
• Short inputs (< 5 words) → always yellow plain text (skip scoring)
• When in doubt, prefer yellow plain text over a wrong category
• Zero friction is the #1 goal — never add steps between input and note creation