TencentAgentMemoryBridge MCP Server

TencentAgentMemoryBridge MCP Server

Provides CodeBuddy, Claude Code, and other MCP-compatible tools with a 4-layer long-term memory system (conversations, atomic facts, scenes, profiles) powered by TencentDB Agent Memory.

Category
Visit Server

README

TencentAgentMemoryBridge

围绕 TencentDB Agent Memory 构建的记忆桥梁生态——将 4 层长期记忆能力(L0 对话 → L1 原子事实 → L2 场景 → L3 画像)接入不同的 Agent 平台。

不造轮子,所有记忆引擎能力由 TencentDB Agent Memory 提供,本仓库仅做协议桥接。

架构

┌─────────────────┐     ┌──────────────┐     ┌──────────────────────────────┐
│  Any MCP Client  │────▶│  MCP Bridge  │────▶│                             │
│ (CodeBuddy,      │     │  (local)     │     │   Bridge Server (public)    │
│  Claude Code,    │     │              │     │   - Auth / Key validation   │
│  Qoder, Reasonix)│     │              │     │   - Sender whitelist        │
└─────────────────┘     └──────────────┘     │   - 按 sender 隔离记忆域     │
                                              │   - Request logging          │
┌─────────────────┐     ┌──────────────┐     │   - HTTP forwarding          │
│  OpenClaw Agent  │────▶│OpenClaw Plugin│    │   POST /api/v1/*            │
│                  │     │(lifecycle hooks)   └──────────┬──────────────────┘
└─────────────────┘     └──────────────┘               │
                                                  ┌──────▼───────────────────┐
                                                  │  TencentDB Agent Memory  │
                                                  │  (upstream, via HTTP)    │
                                                  └──────────────────────────┘

三层

层 位置 职责
Bridge Server 公网服务器 API key 校验 → sender 白名单 → 日志 → 转发到 TencentDB Gateway
MCP Bridge 本地 (MCP clients) MCP 协议 ↔ HTTP 翻译,Sender 按客户端配置
OpenClaw Plugin 本地 (OpenClaw 环境) 生命周期 hooks → HTTP 调用,Sender 在插件配置中

核心原则

Sender 由各 Agent 自行声明,Bridge Server 校验其合法性并记录,但永不修改。每次 recall 自动限定在本 sender 域,确保编码分析和日常对话的记忆互不干扰。

记忆域隔离

每个 sender 拥有独立的记忆域:

Sender 用途 跨域搜索
claude-code 编码分析 ❌
codebuddy 编码调试 ❌
openclaw 日常对话 ✅ 通过复杂命令
  • capture:Bridge Server 自动注入 sender 到转发请求,上游按 sender 标记
  • recall(MCP 工具):自动仅召回本域记忆,工具不暴露 sender 参数
  • recall(HTTP API):支持 sender / senders 参数,跨域需 crossDomainRecall 权限

快速开始

1. 部署 Bridge Server

# 配置环境变量
export PORT=3000
export TENCENTDB_URL=https://your-tencentdb-gateway
export AGENTS='[
  { "name": "claude-code", "apiKeyHash": "<sha256>", "allowedEndpoints": ["recall","capture","search","session"] },
  { "name": "codebuddy", "apiKeyHash": "<sha256>", "allowedEndpoints": ["recall","capture","search"] },
  { "name": "openclaw", "apiKeyHash": "<sha256>", "allowedEndpoints": ["recall","capture"], "crossDomainRecall": true }
]'

# 启动
npx @tencent-agent-memory/bridge-server

2. 配置 MCP Bridge(Claude Code / CodeBuddy)

在 .claude/settings.json 或对应 MCP 客户端配置中注册:

{
  "mcpServers": {
    "agent-memory": {
      "command": "npx",
      "args": ["@tencent-agent-memory/mcp-bridge"],
      "env": {
        "BRIDGE_URL": "https://your-server.com/api/v1",
        "API_KEY": "sk-your-api-key",
        "SENDER": "claude-code"
      }
    }
  }
}

各 Agent 的 SENDER 不同,以此隔离记忆域。详见 Claude Code 配置指南。

3. OpenClaw Plugin

在 OpenClaw 插件配置中填写:

{
  "bridgeUrl": "https://your-server.com/api/v1",
  "apiKey": "sk-your-api-key",
  "sender": "openclaw"
}

插件自动在 before_prompt_build(recall)和 agent_end(capture)等生命周期钩子中桥接记忆。

MCP 工具

工具 端点 说明
recall_memory POST /api/v1/recall 生成前召回相关记忆
store_memory POST /api/v1/capture 生成后存储对话
search_memories POST /api/v1/search/memories 语义搜索记忆
end_session POST /api/v1/session/end 结束当前会话

项目结构

tencent-agent-memory-bridge/
├── packages/
│   ├── bridge-server/        # 公网认证/代理层
│   ├── mcp-bridge/           # MCP Server → HTTP 桥接
│   └── openclaw-plugin/      # OpenClaw 生命周期 hooks
├── examples/
│   ├── codebuddy/            # CodeBuddy 配置示例
│   └── claude-code/          # Claude Code 配置指南
├── docs/
│   └── design-overview.md    # 完整架构文档 (v0.4-draft)
├── CLAUDE.md                 # 项目指令 + Auto Memory Store 规则
└── package.json

上游依赖

License

MIT

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured