---

name: feature-reminders description: Remind about underutilized Claude Code features at appropriate times. Use when errors occur, context is high, or searches are large. Keywords: rewind, context, compact, explore, tip, reminder, help allowed-tools: Read

Feature Reminders

Purpose

This skill reminds Claude about powerful features at the right moments, so you don't have to remember them.

When to Suggest /rewind

Triggers:

• After a tool error or failure
• When the same approach failed 2+ times
• When code changes need to be undone

Reminder:

💡 /rewind can rollback both code AND conversation to a previous checkpoint. Press Esc+Esc as a shortcut.

When to Suggest /compact

Triggers:

• Memory entity count > 60
• Estimated tokens > 8,000
• Session feels sluggish
• Context window visualization shows high usage

Reminder:

💡 /compact [instructions] can optimize memory. Give it focus instructions like "keep bug fixes, archive explorations".

When to Suggest /context

Triggers:

• After 3+ edits without checking context
• When token usage is unclear
• When unsure what Claude "remembers"

Reminder:

💡 /context shows a visual grid of how the context window is being used. Helps identify what's consuming tokens.

When to Suggest Explore Subagent

Triggers:

• Recursive glob patterns (**)
• Complex regex searches
• "Find all" or "search everywhere" requests
• Architecture questions

Reminder:

💡 Use Task with subagent_type: Explore for large codebase searches. It's Haiku-powered and saves your context.

Keyboard Shortcuts to Remember

Self-Check Questions

Before making changes, Claude should ask itself:

1. Did I check /context recently?
2. Is memory above 60 entities?
3. Should I use Explore subagent instead of direct search?
4. If this fails, can I /rewind?