Retrieve historical AI session data via semantic vector and keyword search.
npx skills add https://github.com/cwinvestments/memstack --skill echo-cwinvestments
Please help me install this Agent Skill. Skill: echo Source: https://github.com/cwinvestments/memstack/tree/main/skills/echo Command: npx skills add https://github.com/cwinvestments/memstack --skill echo-cwinvestments
This Skill retrieves information from past AI sessions, preventing the need to manually search through old logs or rely on fragmented memory.
Use the echo skill to search for information about the 'user authentication' feature from past sessions.
High-intent search queries and answers about installing and using this skill.
To recall past AI session history, this Skill retrieves past interactions using semantic vector search and keyword-based database queries. It combines LanceDB vector embeddings and SQLite structured data to find exact log entries from previous sessions.
Yes, you can search past session context using semantic vector search. This Skill leverages LanceDB to find relevant past conversations based on the meaning of your query, ensuring comprehensive recall across multiple sessions and projects.
To retrieve specific decisions and tasks from previous project sessions, query the stored historical session data. This Skill supports recall of accomplished tasks and pending items across multiple projects using SQLite and markdown file fallbacks.
Yes, this session recall tool works with LanceDB and SQLite. It leverages LanceDB for vector embeddings and SQLite for structured data queries, while also using markdown files as a fallback for comprehensive historical data retrieval.
The best way to search historical AI interactions across multiple projects is using a multi-source recall approach. This Skill combines vector database results, SQLite queries, and markdown file fallbacks to find relevant past conversations based on semantic meaning.
Using markdown files for session memory recall is limited to a fallback mechanism. The primary recall system leverages LanceDB vector embeddings and SQLite structured data, meaning markdown fallbacks may lack the semantic search depth of the database queries.
nimadorostkarStore agent state, architectural facts, and task outcomes in Markdown files.
troykellyAggregate project and task data to generate structured status reports.
hiyenwongStore and search chat histories locally with LanceDB vector embeddings and RAG retrieval.
manifoldlogicConvert code chunks into numerical vectors for semantic codebase search.
