Recall relevant memories from long-term storage using semantic search
Recalls relevant memories from long-term storage using semantic search and metadata filters.
/plugin marketplace add greyhaven-ai/claude-code-config/plugin install knowledge-base@grey-haven-plugins<query>Recall memories: $ARGUMENTS
<ultrathink>
The past informs the present. Memories recalled at the right moment become superpowers for decision-making.
</ultrathink>
<megaexpertise type="memory-architect">
The assistant should use the memory-architect agent to perform semantic search across stored memories, returning the most relevant historical context.
</megaexpertise>
<context>
Query: $ARGUMENTS
Memory store path: .claude/memory/
Current directory: !pwd
Search types available: semantic (vector), full-text (BM25), metadata (filters)
</context>
<requirements>
Launch Memory Architect Agent:
Use the Task tool with subagent_type="general-purpose" to invoke the memory-architect agent.
Prompt:
"Search long-term memory for: $ARGUMENTS
Please:
1. Check if ContextFrame is installed and memory store exists
2. If available, perform semantic search using vector embeddings
3. Otherwise, use BM25 full-text search
4. Apply any metadata filters (type, tags, status, date range)
5. Rank results by relevance score
6. For top 5 results, show:
- Title/summary (first 2 lines)
- Memory type and date
- Tags and components
- Relevance score
- UUID for retrieval
7. Provide full content for the most relevant result
8. Suggest related memories or knowledge entries
If ContextFrame is not installed, provide instructions for installation."
Present Results:
The assistant should make memory recall feel like having a knowledgeable colleague who remembers everything, surfacing the most relevant historical context instantly.