wtf is pliamem-public?

Pliamem is a memory microservice for AI agents that addresses memory fragmentation, the problem of an AI agent needing context spread across many disconnected sources at once, such as a vector database, a knowledge graph, session logs, documentation files, and cloud storage. Instead of querying each source separately and combining results by hand, Pliamem works as a single routing layer. You send it one query, and it searches every configured memory layer at the same time, scores the results by recency and relevance, removes duplicates, and returns context already formatted to fit within an AI model's token limit. It supports five memory layer types out of the box: a semantic vector search layer for Python based OMPA brains, a knowledge graph layer for structured entity and relationship data, a flat file layer for Markdown documentation, a chronological daily log layer, and a cloud key value storage layer backed by a service called Puter. From the command line, pliamem chat starts an interactive AI conversation that automatically pulls in relevant memory context with each message. pliamem search runs a one off query with a configurable token limit. pliamem sync pushes local files and logs to the cloud key value store. pliamem serve starts a REST API server so other agents or tools can query the memory system remotely. A Vite based web dashboard gives a visual overview of the memory stack. The library is built in Node.js following an Adapter design pattern, so new storage backends can be added without changing the core routing logic. It is installable globally through NPM and also listed on PyPI. The project is copyrighted by Micap AI LLC with all rights reserved, rather than released under a standard open source license, so reuse and redistribution terms are not the typical permissive or copyleft arrangement.