remember

/remember [type]

Save something worth keeping - a fix that worked, a pattern you want to reuse, a decision you made, or a pitfall to avoid.

Argument

• type (optional): solution, pattern, decision, gotcha

Workflow

Step 1: Determine Type

If no type was provided, ask:

What do you want to save?
1. solution - An error you fixed (problem + root cause + fix)
2. pattern  - An approach that worked well
3. decision - An architecture or design choice you made
4. gotcha   - A pitfall or trap to avoid next time

Step 2: Ask Type-Specific Questions

Keep this conversational. Ask all questions for the type at once, not one by one.

Solution:

• What was the problem or error?
• What was the root cause?
• What fixed it?

Pattern:

• What is this pattern called?
• When should it be used?
• How is it applied?

Decision:

• What question was decided?
• What was decided, and why?
• What alternatives were rejected?

Gotcha:

• What is the pitfall?
• Why is it a problem?
• What is the correct approach?

Step 3: Generate Tags

Generate 3-5 tags based on:

• The active project (if known from .claude/memory/index.json)
• Technology or tool mentioned
• Keywords from the description

Format tags as lowercase slugs: next-js, api-auth, caching

Step 4: Save the Learning

Create the file at:

.claude/memory/learnings/{type}-{YYYY-MM-DD}-{short-slug}.md

Where {short-slug} is a 2-3 word kebab-case summary of the content.

File format:

---
date: {YYYY-MM-DD}
type: {type}
tags: [{tag1}, {tag2}, {tag3}]
project: {active project name, or "general" if unknown}
---

# {Title summarizing what was learned}

## Context

{Brief description of where/when this came up}

## {Type-specific content}

{The actual content based on what was answered in Step 2}

For each type, use these section headers inside the file:

• solution: ## Problem, ## Root Cause, ## Fix
• pattern: ## When to Use, ## How to Apply
• decision: ## Decision, ## Reasoning, ## Rejected Alternatives
• gotcha: ## The Pitfall, ## Why It's a Problem, ## Correct Approach