Searches and retrieves historical discussions, design decisions, and domain knowledge from a Supabase-backed long-term memory system.
This skill empowers Claude with long-term persistent memory by providing a structured way to query a Supabase 'memories' database. It allows the agent to recall past architectural decisions, technical discussions, and specialized domain knowledge that occurred in previous sessions. By leveraging multiple SQL search patterns—including keyword matching, category filtering, and vector similarity—it ensures that relevant context is always available, maintaining project continuity and preventing the loss of critical insights over time.
主な機能
01Importance-based ranking to surface critical information first
020 GitHub stars
03Flexible SQL-based searching via Supabase MCP
04Multi-mode retrieval including keyword, category, and tag filtering
05Context-efficient architecture that minimizes token usage
06Support for vector similarity search for semantic context retrieval
ユースケース
01Recalling specific architectural decisions from past project phases
02Reviewing historical team discussions to understand legacy code rationale
03Searching for specialized project-specific knowledge and best practices
