harrylettering/codememory/codememory-context-skill

CodeMemory Context Skill

Automatically retrieves relevant conversation history when the user asks about past discussions.

When This Skill Applies

This skill activates when the user's request involves:

• Asking about "what we discussed earlier"
• Referencing "previous" decisions or conversations
• Questions like "do you remember when..."
• Needing context from earlier in the session or past sessions
• Any query that would benefit from historical context

How to Use This Skill

When activated:

1. Identify the Context Need: Determine what time period or topic the user is asking about
2. Retrieve Relevant Content: Use /codememory-grep to search for matching content
3. Expand if Needed: Use /codememory-expand on promising summaries for more detail
4. Inject Context: Add the retrieved information to the conversation context
5. Answer the User: Use the injected context to provide an informed answer

Context Injection Format

When adding retrieved context, format it as:

--- BEGIN CODEMEMORY CONTEXT ---
 
## Retrieved from Conversation History:
 
[Summary: sum_abc123 (2026-04-04)]
Key discussion points about database schema...
 
[Message: msg_456 (2026-04-04, USER)]
"We should use SQLite with FTS5 for full-text search"
 
--- END CODEMEMORY CONTEXT ---

Best Practices

• Be selective: Only retrieve the most relevant context to avoid token bloat
• Cite sources: Include summary/message IDs so user can verify
• Ask for clarification: If the query is ambiguous about what time period to search
• Use sub-agent: For complex queries, delegate to codememory-query-agent

Tools Available

• /codememory-grep - Search conversation history
• /codememory-describe - Get details about a specific summary
• /codememory-expand - Expand a summary for more detail
• /codememory-expand-query - Ask a question with sub-agent delegation
• /codememory-status - Check system status