Recall

Search memory for relevant learnings, decisions, and patterns.

Usage

/recall authentication
/recall "how did we fix the race condition"
/recall error handling patterns

How It Works

1. Takes your query
2. Searches the memory database (text + optional semantic search)
3. Returns the most relevant learnings
4. Shows type, content, confidence, and when it was stored

Execute

Run the recall script:

"${CLAUDE_PLUGIN_ROOT}/scripts/memory.sh" recall --query "$ARGUMENTS" --limit 5

Output Format

Present results clearly:

## Memory Recall: "authentication"

### 1. [WORKING_SOLUTION] High Confidence
JWT auth requires checking token expiry before validation
*Stored: 2024-01-10*

### 2. [ARCHITECTURAL_DECISION] High Confidence
Using refresh tokens with 7-day expiry, access tokens 15 min
*Stored: 2024-01-08*

### 3. [FAILED_APPROACH] Medium Confidence
Don't store tokens in localStorage - use httpOnly cookies
*Stored: 2024-01-05*

Tips

• Be specific: "auth token validation" better than just "auth"
• Include context: "database connection pooling in Python"
• Ask questions: "how did we handle rate limiting"

If no results found, suggest checking different terms or offer to search more broadly.