OMOS-MCP
OMOS-MCP
MCP 服务器让 AI 代理(Claude、ChatGPT、Cursor 等)读取 OMOS Shared Drive 中的项目文档来回答问题 — 每个答案都引用源文件并附带 Google Drive 链接
支持的 Drive 结构:
OMOS/
├── Project A/
│ ├── Project Overview/
│ ├── Timeline/
│ ├── Design/
│ │ ├── System Flow/
│ │ ├── DB/
│ │ └── API/
│ └── BRD/
└── Project B/ ...工具
工具 | 说明 |
| 索引整个 Drive:所有项目 + 所有文件,附带 ID 和链接(代理始终先调用此工具) |
| 全文搜索整个 Drive,可按项目/部分限定范围 |
| 以文本形式读取文件 — Google Docs/Sheets/Slides、PDF、.docx、.xlsx、md/text 以及图片(直接将图片发送给代理查看) |
| 立即重建索引(默认缓存 5 分钟) |
代理按顺序搜索:过滤 Project Name → 未找到则查看 Project Overview 并向用户确认 → 继续询问是否涉及 Timeline | BRD | Design (System Flow / DB / API) → 问题过于宽泛时会反问以缩小范围(已内置于服务器指令中)
Related MCP server: brkt-gdrive-mcp
一次性设置:Service Account
服务器通过 service account 访问 Drive(无需每个人登录):
进入 console.cloud.google.com → 创建新项目(或使用现有项目)
APIs & Services → Library → 搜索 Google Drive API → 点击 Enable
IAM & Admin → Service Accounts → Create Service Account → 设置名称如
omos-mcp→ 点击 Create(可跳过角色步骤,无需授予任何权限)进入创建的 service account → 标签页 Keys → Add Key → Create new key → JSON → 密钥文件将下载
复制 service account 的电子邮件(格式为
omos-mcp@<project>.iam.gserviceaccount.com)并共享 Shared Drive / OMOS 文件夹,授予该邮箱 Viewer 权限找到根 OMOS 的 folder id:在浏览器中打开文件夹,查看 URL
https://drive.google.com/drive/folders/<这是 id>
环境变量
变量 | 说明 |
| root OMOS 的文件夹 ID(来自步骤 6)— 必填 |
| 整个 JSON 密钥文件的内容,或指向文件的路径 — 必填 |
| (HTTP 模式)客户端必须发送的 bearer token |
| 索引缓存过期时间(秒),默认 300 |
| (HTTP 模式)为 Claude.ai / ChatGPT 网页版启用 OAuth 模式 |
| (HTTP 模式)监听端口,默认 8000 |
本地运行
先安装 uv(curl -LsSf https://astral.sh/uv/install.sh | sh)
1) 设置 .env(服务器从仓库根目录自动加载):
cp .env.example .env编辑 .env,填入 OMOS_ROOT_FOLDER_ID 和密钥文件路径(例如将密钥文件放在 ./service-account.json — 已加入 gitignore)
2) 连接 Claude Code(stdio):
claude mcp add omos -- uv run --directory "/path/to/OMOS-MCP" omos-mcpClaude Desktop / Cursor / Windsurf — 在 MCP 配置中添加:
{
"mcpServers": {
"omos": {
"command": "uv",
"args": ["run", "--directory", "/path/to/OMOS-MCP", "omos-mcp"]
}
}
}3) 测试: 向代理提问 "OMOS 中有哪些项目?" — 应返回项目列表及链接
(可选)在本地作为 HTTP 服务器运行: 在 .env 中启用 OMOS_AUTH_TOKEN,然后
uv run omos-mcp-http端点位于 http://localhost:8000/mcp,连接方式:
claude mcp add --transport http omos http://localhost:8000/mcp -H "Authorization: Bearer <OMOS_AUTH_TOKEN>"部署为 MCP link(Render)
将此仓库推送到 GitHub(已包含
Dockerfile+render.yaml)Render → New → Blueprint → 选择此仓库
在仪表盘中设置环境变量:
OMOS_ROOT_FOLDER_ID= root OMOS 的文件夹 IDGOOGLE_SERVICE_ACCOUNT_JSON= 整个 JSON 密钥文件的内容(直接粘贴)OMOS_AUTH_TOKENRender 自动生成 → 复制分发给团队
部署完成后 → 端点为
https://<app>.onrender.com/mcp
用户连接:
claude mcp add --transport http omos https://<app>.onrender.com/mcp \
-H "Authorization: Bearer <OMOS_AUTH_TOKEN>"Cursor / VS Code:在 MCP 配置中添加 URL 和 header Authorization: Bearer <token>
供 Claude.ai / ChatGPT(网页版)使用 — OAuth
网页客户端没有输入 header 的地方,必须通过 OAuth 提供商(如 WorkOS AuthKit,免费支持最多 100 万用户):
注册 workos.com → 启用 AuthKit → 设置登录方式并限制仅团队成员
仪表盘 → Connect → Configuration → 开启 Client ID Metadata Document 和 Dynamic Client Registration
添加 Resource Indicator =
https://<app>.onrender.com(不带/mcp)在 Render 上额外设置 3 个环境变量:
OAUTH_ISSUER= AuthKit 域名,OAUTH_AUDIENCE=https://<app>.onrender.com,PUBLIC_URL=https://<app>.onrender.com用户:Claude.ai → Settings → Connectors → Add custom connector → 输入
https://<app>.onrender.com/mcp→ 通过 WorkOS 登录
OAuth 模式和 bearer token 可以共存 — 网页使用 OAuth,CLI 仍可使用 token
测试转换器
uv run python test_convert.pyThis server cannot be installed
Maintenance
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
- Flicense-qualityDmaintenanceAn MCP server that enables users to query documents from Google Drive using AI assistance, providing tools for searching vector databases and generating answers based on document content.
- Flicense-qualityBmaintenanceMCP server for semantic search over Google Drive documents, enabling AI tools to search, list, and retrieve document content.
- Flicense-qualityDmaintenanceAn MCP server that enables AI assistants to interact with Google Docs, Sheets, and Drive using the user's own Google account.
- Alicense-qualityDmaintenanceMCP server that reads Google Docs and Google Sheets content for AI assistants like Claude.111MIT
Related MCP Connectors
MCP server for generating rough-draft project plans from natural-language prompts.
Driflyte MCP server which lets AI assistants query topic-specific knowledge from web and GitHub.
MCP server for AgentDocs (agentdocs.eu): read, search, write, comment on & share Markdown docs.
Latest Blog Posts
- Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)By Om-Shree-0709 on .mcpAgent IdentityOAuth 2.1
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
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/Pd011161/OMOS-MCP'
If you have feedback or need assistance with the MCP directory API, please join our Discord server