Claude Code Plugins

Community-maintained marketplace

Feedback

>

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name stash-pad
description Use this skill when the user wants to capture todos, ideas, or thoughts quickly during CLI coding sessions. Activate when user says :add, :a, :done, :d, :classify, :c, :show, :s, :now, :n, :restore, :r, :find, :f, :archive, or :ar. Manages a TODO.md file with inbox, categories, completion tracking, and restore. Works mid-session without disrupting workflow.
allowed-tools Read, Edit, Write

Stash Pad

A universal skill for capturing, organizing, and classifying thoughts, ideas, and todos using a simple file-based inbox system.

Overview

This skill enables any LLM assistant to help users capture thoughts quickly and organize them later. It follows a "capture first, classify later" workflow that reduces friction when brainstorming.


Triggers

Activate this skill when the user says any of the following:

Command Shortcut Action
:add :a Add item(s) to inbox
:done :d Mark item complete
:classify :c Organize inbox
:show :s Display list
:now :n Add + classify
:restore :r Restore from completed
:find :f Search all todos
:archive :ar Clear completed items

Tip: To customize triggers, edit this file directly.

Activation Guidelines

Use this skill when:

  • User explicitly wants to manage todos/tasks
  • Command appears at the start of user input
  • Context suggests todo management intent

Do NOT activate when:

  • :a, :d, etc. appear mid-sentence or in code/paths
  • User is discussing something unrelated to task management
  • The colon-prefixed text is part of a filename, URL, or code snippet

Core Behaviors

1. Capturing Items

When the user provides items to capture:

  1. Parse the input:

    • If input contains semicolons (;), split into multiple items
    • Trim whitespace from each item
    • Preserve the user's original wording
  2. Add to Inbox:

    • Open the project's TODO.md file
    • If ## Inbox section doesn't exist, append it to the end of the file first
    • Append each item to the ## Inbox section as - [ ] <item>
    • Confirm what was added
  3. Suggest classification (optional):

    • Briefly note what category each item might belong to
    • Example: "Added. (That's likely a Feature)"

2. Classifying Items

When the user says "classify" or similar:

  1. Read the Inbox section of TODO.md

  2. For each item, determine the best category:

    • Features - New functionality
    • Improvements - Enhancements to existing functionality
    • Bug Fixes - Issues that need fixing
    • Documentation - Docs, README, examples
    • Ideas - Exploratory thoughts, research topics
  3. Group related items under subheadings when patterns emerge

    • Example: Multiple UI items → create "### User Interfaces" subheading
  4. Move items from Inbox to their categories

  5. Clear the Inbox after classification

  6. Show summary of what was classified where

  7. Handle empty inbox:

    • If inbox is empty, respond: "Inbox is empty. Nothing to classify."

3. Completing Items

When the user says :done <text> or :d <text>:

  1. Search all categories in TODO.md for matching items

    • Match partial text (case-insensitive)
    • Example: done: auth matches - [ ] add user authentication
  2. For each match:

    • Change - [ ] to - [x]
    • Move the item to ## Completed section
    • Add date prefix: - [x] 2024.11.30 - item text
  3. Handle ambiguity:

    • If multiple items match, list them and ask which one(s)
    • If no items match, respond: "No matching items found for ''."
  4. Confirm completion:

    • Brief confirmation with the item text
    • Example: ✓ Marked "add user authentication" complete

4. Add and Classify

When the user says :now <text> or :n <text>:

  1. Add item(s) to Inbox (same as :add)
  2. Immediately classify the new item(s)
  3. Return combined confirmation:
    • Example: Added and classified: "add dark mode" → Features

5. Restoring Items

When the user says :restore or :r:

  1. Without arguments - Display completed items (newest first):

    Completed (newest first):
    1. 2024.11.30 - add user authentication
    2. 2024.11.29 - fix login bug
    3. 2024.11.28 - update README
    
    Restore with :r <number> or :r <text>
    
  2. With number(s) - :r 1 or :r 1 3:

    • Restore specified items to Inbox
    • Remove from Completed section
  3. With text - :r auth:

    • Search Completed for matches
    • Handle ambiguity same as :done
  4. Confirm restoration:

    • Example: Restored "add user authentication" to inbox.

6. Searching Items

When the user says :find <text> or :f <text>:

  1. Search all sections of TODO.md for matching items

    • Case-insensitive partial match
  2. Return matches with location:

    Found 3 matches for "auth":
    
    Features:
    - add user authentication
    
    Bug Fixes:
    - fix auth token expiry
    
    Completed:
    - 2024.11.28 - remove old auth code
    
  3. Handle no matches:

    • Respond: "No items found matching ''."

7. Archiving Completed Items

When the user says :archive or :ar:

  1. Clear the Completed section:

    • Remove all items from ## Completed
    • Or move to ## Archived section if it exists
  2. Confirm action:

    • Example: Archived 5 completed items.
  3. Handle empty:

    • If Completed is empty, respond: "Nothing to archive."

8. Creating TODO.md

If TODO.md doesn't exist:

  1. Ask the user first: "No TODO.md found. Create one?"
  2. If user confirms, create it with this template:
# Project Todo List

## Features
<!-- New functionality to add -->

## Improvements
<!-- Enhancements to existing functionality -->

## Bug Fixes
<!-- Issues that need fixing -->

## Documentation
<!-- Docs, README, examples -->

## Ideas
<!-- Raw thoughts to classify later -->

---

## Inbox
<!-- Drop any thought here - assistant will classify it -->

---

## Completed
<!-- Done items moved here automatically -->

Response Style

  • Be brief when adding items - just confirm what was added
  • Be helpful when classifying - show the organization clearly
  • Use tables or trees to display structure when helpful
  • Don't over-explain - the user knows what they meant

Good Response (Adding)

Added 3 items to inbox.

Good Response (Classifying)

Classified 5 items:

Features (3)
├── User authentication
├── Dark mode
└── Export to PDF

Improvements (2)
├── Faster load times
└── Better error messages

Inbox is empty.

Customization

Project-Specific Categories

Users may have custom categories. Detect these from the existing TODO.md structure and use them. Common variations:

  • ## Technical Debt - Code that needs refactoring
  • ## Research - Topics to investigate
  • ## Questions - Things to clarify with stakeholders
  • ## Blocked - Items waiting on dependencies

Priority Markers

If the user includes priority markers, preserve them:

  • [P0], [P1], [P2] - Priority levels
  • [urgent], [low] - Priority labels
  • !!! - High priority indicator

Examples

Adding Items

User: :a add user authentication Assistant: Added to inbox.

User: :add fix login bug; add dark mode; update README Assistant: Added 3 items to inbox.

Classification

User: :c Assistant: Classified 3 items:

Item Category
Fix login bug Bug Fixes
Add dark mode Features
Update README Documentation

Ambiguous Match

User: :done bug Assistant: Found 2 matches:

  1. Fix login bug
  2. Fix signup bug

Which one(s)? (1, 2, or both)

Restore

User: :r Assistant: Completed (newest first):

  1. 2024.11.30 - add user authentication
  2. 2024.11.29 - fix login bug

Restore with :r or :r

User: :r 1 Assistant: Restored "add user authentication" to inbox.


File Location

By default, look for TODO.md in the project root. If not found:

  1. Ask user if they want to create one
  2. Or check for alternatives: TODOS.md, todo.md, .todo/inbox.md

Cross-LLM Compatibility

This skill is designed to work with any LLM that can:

  • Read and write files
  • Parse markdown
  • Follow instruction patterns

Claude Code

Place this file in .claude/skills/ or reference it in CLAUDE.md

Gemini

Include these instructions in your system prompt or context

OpenAI Codex

Add to your instructions file or system context

Other LLMs

Adapt the triggers and file operations to your platform's capabilities


Version

2.0.1 - Anthropic guidelines compliance; activation guardrails; user consent for file creation 2.0.0 - Added restore, find, archive commands; date-prefixed completed items 1.0.0 - Initial release