Gomaa — 自主代理内存操作系统。面向AI代理的持久内存系统，支持Obsidian保险库集成、混合RRF搜索、知识图谱、安全门和MCP服务器。
概览
Gomaa为AI代理（Hermes、OpenClaw、Claude Desktop、Cursor、Windsurf、CrewAI、LangChain）配备了永久的结构化长期记忆。它连接了人类可读与高速或零配置，支持混合互惠等级融合（RRF）搜索、维基链接知识图谱、Ebbinghaus时间衰减、跨代理舰队共享以及异步Google Drive云同步。 大多数人工智能记忆系统存在三个根本缺陷：1. 记忆会消失在不透明的矢量数据库中。人类无法审计、修正或整理智能体所学内容。2. 如果不忘记机制，旧噪音会积累并污染智能体的提示窗口。3. 研究笔记、资历和任务笔记本相互碰撞，导致幻觉。 * 📖 每段记忆都是你Obsidian保险库中人类可读的Markdown笔记，配有[[Wiki链接]]和YAML前言。* ⏳ 不活跃的记忆呈指数级消退（$Salience \times 0.95^{\Delta t}$），而 #pinned 记忆则永久存在。
概述（AI 翻译）
生产级、本地优先的分层式和神经启发的自主AI智能体认知记忆引擎（v4.0.0 — “Gomaa Neo”）。
Gomaa为AI代理（Hermes、OpenClaw、Claude Desktop、Cursor、Windsurf、CrewAI、LangChain）配备了永久、类人且超人类的长期记忆。它结合了人类可读的Obsidian Markdown Vaults与高速PostgreSQL + pgvector（HNSW）或零配置SQLite WAL，支持4象限认知架构（情节、语义、程序、工作记忆）、**HippoRAG 2个性化页面排名（PPR）**关联检索、三阶段生物睡眠巩固（SWS + REM + 突触稳态）、ACT-R基础级激活、混合互惠等级融合（RRF）搜索、Ebbinghaus时间衰减、跨代理舰队共享及异步Google Drive云同步。
💡 为什么是Gomaa？
大多数人工智能记忆系统存在三个根本性缺陷：
- 黑匣子矢量斑点： 记忆会消失在不透明的矢量数据库中。人类无法审计、纠正或整理代理人所学到的内容。
本页由 AI 翻译，技术细节请以原文为准。
README
以下为原文 README 节选，完整内容请查看原文。
Gomaa 🧠
Production-grade, local-first hierarchical & neuro-inspired cognitive memory engine for autonomous AI agents (v4.0.0 — “Gomaa Neo”).
Gomaa equips AI agents (Hermes, OpenClaw, Claude Desktop, Cursor, Windsurf, CrewAI, LangChain) with permanent, human-like and transhuman long-term memory. It unites human-readable Obsidian Markdown Vaults with high-speed PostgreSQL + pgvector (HNSW) or zero-config SQLite WAL, powering a 4-Quadrant Cognitive Architecture (Episodic, Semantic, Procedural, Working Memory), HippoRAG 2 Personalized PageRank (PPR) associative retrieval, 3-Phase Biological Sleep Consolidation (SWS + REM + Synaptic Homeostasis), ACT-R Base-Level Activation, hybrid Reciprocal Rank Fusion (RRF) search, Ebbinghaus temporal decay, cross-agent fleet sharing, and asynchronous Google Drive cloud synchronization.
💡 Why Gomaa?
Most AI memory systems suffer from three fundamental flaws:
- Black-Box Vector Blobs: Memories disappear into opaque vector databases. Humans cannot audit, correct, or curate what the agent learned.
- Context Pollution & Flat Storage: Raw chat logs, factual knowledge, and procedural skills are dumped into a single flat vector index without forgetting mechanisms or cognitive separation.
- Multi-Hop Blindness & Domain Cross-Contamination: Standard vector similarity fails when connecting facts across multiple hops, while un-scoped stores leak context across domains.
Gomaa v4.0 (“Gomaa Neo”) solves this:
- 🧠 4-Quadrant Cognitive Architecture: Explicit separation of Episodic (
episodic_eventsturn ledger), Semantic (notes+ weightedlinks), Procedural (procedural_playbookswith Bayesian Laplace utility), and Working Memory (working_memorybounded by Miller’s Law $7 \pm 2$). - 🕸️ HippoRAG 2 Associative PPR Retrieval: Single-step multi-hop retrieval using Personalized PageRank ($K=20, \alpha=0.15$) over bi-directional weighted graph links, fused with dense vectors, BM25 FTS, and ACT-R base-level activation ($\ln(1 + \text{access_count})$).
安装
This server does not publish a one-line install command.
Open the repository installation guide 新手？先看 mcpskill.org 教程 →配置
{
"mcpServers": {
"gomaa": {
"command": "python3",
"args": ["-m", "gomaa", "server"],
"env": {
"MEMORY_VAULT_PATH": "~/.gomaa/vault",
"MEMORY_DEFAULT_WING": "general"
}
}
}
}