Local Agent Memory Without Docker: amem + MCP

Описание и применение

amem — это локальный слой памяти для AI-агентов через MCP, хранящий факты в ~/.amem без необходимости использования Docker или облачных сервисов.

Задача

Coding agents forget context between chats, requiring users to re-explain preferences, decisions, and tool connections in every new thread, or rely on cloud memory products.

Результат

A local, Docker-free memory solution for AI agents that stores facts in ~/.amem, enabling persistent context across sessions without cloud dependency.

Рабочий процесс

• Install amem using npx @iamem/amem setup or globally via npm i -g @iamem/amem && amem setup.
• Launch the local UI using amem ui or amem app.
• Configure the agent to use amem_context to pull compact relevant facts before exploring or prompting.
• Use amem_remember to store one durable sentence after a real outcome, such as a preference, decision, or recipe.

Доказательства

Facts live in ~/.amem on your machine. Free to start. No telemetry.

Источник

amem_context — pull compact relevant facts before exploring or prompting

Источник