Skip to main content
Glama

OpenCode Memory System

基于 MCP 协议的 AI 记忆系统,支持语义向量检索、双记忆引擎、Git 版本管理。可接入任何支持 MCP 的 AI 客户端(OpenCode、HERMES、Claude Desktop、Cursor 等)。

功能特性

功能

说明

语义向量检索

基于 sentence-transformers + FAISS,支持中英文语义搜索

双记忆引擎

短期记忆(JSON 文件)+ 长期记忆(Markdown)

Git 版本管理

记忆自动 commit,支持回滚

自动编码识别

UTF-8 优先,自动回退系统编码,兼容 Windows 中文环境

代理字符防护

递归清洗 UTF-16 代理字符,防止编码崩溃

MCP 标准协议

JSON-RPC over stdio,任何 MCP 客户端可直接调用

Related MCP server: knowledgeplane

提供的 MCP 工具

工具名

说明

memory_remember

保存记忆(自动存储 + Git commit)

memory_recall

语义搜索记忆(合并 FAISS 索引 + 实时文件搜索)

memory_forget

按关键词删除记忆

memory_status

查看系统状态(索引、模型、记忆数量)

memory_reindex

重建向量索引(支持后台运行)

memory_history

查看 Git 变更历史

memory_rollback

回滚到指定版本

memory_sync

初始化/同步 Git 仓库

快速安装

1. 安装依赖

pip install -r requirements.txt

2. 下载模型(首次运行自动下载)

python scripts/semantic_search.py --download

3. 配置 MCP 客户端

OpenCode (~/.config/opencode/opencode.jsonc)

{
  "mcp": {
    "memory": {
      "type": "local",
      "command": ["python", "-u", "scripts/mcp_server.py"],
      "enabled": true
    }
  }
}

HERMES (~/.hermes/config.yaml)

mcp_servers:
  memory:
    type: stdio
    command: ["python", "-u", "/path/to/scripts/mcp_server.py"]

Claude Desktop (claude_desktop_config.json)

{
  "mcpServers": {
    "memory": {
      "command": "python",
      "args": ["-u", "/path/to/scripts/mcp_server.py"]
    }
  }
}

目录结构

opencode-memory/
├── scripts/                    # MCP Server 核心脚本
│   ├── mcp_server.py          # MCP Server 主入口(JSON-RPC over stdio)
│   ├── semantic_search.py     # 语义向量检索(sentence-transformers + FAISS)
│   ├── dual_memory_engine.py  # 双记忆协同引擎
│   ├── build_full_index.py    # 向量索引构建工具
│   ├── memory_config.py       # 路径配置(自动适配系统)
│   ├── memory_git.py          # Git 自动版本管理
│   ├── storage_tiers.py       # 三层存储(热/温/冷)
│   ├── memory_maintain.py     # 记忆维护工具
│   ├── memory_dedup.py        # 记忆去重
│   ├── auto_index_updater.py  # 自动索引更新
│   ├── plugin_bridge.py       # 插件桥接(BM25 混合检索)
│   └── init_dual_memory.py    # 双记忆引擎初始化
├── plugins/
│   └── memory-plugin/
│       └── index.ts           # OpenCode 插件(可选)
├── skills/
│   └── memory-system/
│       └── SKILL.md           # AI 技能文档
├── requirements.txt           # Python 依赖
├── .gitignore
└── README.md

工作原理

编码处理流程

客户端发送 UTF-8 字节
        ↓
┌─────────────────────────────────────┐
│  _detect_decode() 自动识别编码       │
│  ① 先试 UTF-8(MCP 标准协议)        │
│  ② 失败→试系统编码(cp936/cp1252)   │
│  ③ 再失败→UTF-8 + 替换非法字符       │
└─────────────────────────────────────┘
        ↓
┌─────────────────────────────────────┐
│  _clean_surrogates() 移除代理字符    │
│  _clean_obj() 递归清洗整个响应       │
└─────────────────────────────────────┘
        ↓
  存储为 UTF-8 JSON 文件

记忆检索流程

memory_recall("查询")
        ↓
   ┌────┴────┐
   ↓         ↓
FAISS 索引  STM 文件实时编码
(预构建)    (即时搜索)
   ↓         ↓
   └────┬────┘
        ↓
   合并去重 + 按相似度排序
        ↓
   返回 Top-K 结果

依赖

  • Python 3.10+

  • sentence-transformers(语义编码)

  • faiss-cpu(向量索引)

  • numpy

许可证

MIT License

F
license - not found
-
quality - not tested
B
maintenance

Maintenance

Maintainers
Response time
Release cycle
Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Servers

  • A
    license
    -
    quality
    C
    maintenance
    An MCP server that provides persistent, cross-session memory and team knowledge sharing for AI development workflows. It enables project DNA scanning, semantic search, context budgeting, and git-aware indexing to prevent AI context loss between sessions.
    Last updated
    30
    MIT
  • F
    license
    -
    quality
    B
    maintenance
    MCP server that gives AI agents and teams persistent, shared memory using a knowledge graph with vector embeddings, automatic consolidation of related facts, and hybrid search.
    Last updated
    3
  • A
    license
    -
    quality
    B
    maintenance
    A self-hosted MCP server that provides a personal semantic memory layer for AI tools. It enables storing, searching, and managing memories using hybrid vector and keyword search, allowing AI assistants to recall information by meaning.
    Last updated
    MIT

View all related MCP servers

Related MCP Connectors

  • Cloud-hosted MCP server for durable AI memory

  • Person-owned, portable AI memory as a remote MCP server, readable and writable by any MCP client.

  • User-owned memory for AI agents, Copilot, Claude, IDEs, CLIs, and chat apps over remote MCP.

View all MCP Connectors

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Shine8592/opencode-memory'

If you have feedback or need assistance with the MCP directory API, please join our Discord server