mcp-viva-engage
Enables Claude AI to interact with Viva Engage (Yammer) for browsing communities, searching posts, reading conversations, and posting replies.
README
mcp-viva-engage
An open source Model Context Protocol (MCP) server that connects Claude AI to Viva Engage (Yammer). Browse your home network communities, search posts, read conversations, and post replies — all through natural conversation with Claude.
Built by Pranav Joshi
What You Can Do
Once connected, you can ask Claude things like:
- "Summarize what's new in all my communities today"
- "What is new in ISV Collaboration this week?"
- "Find all posts about AL Extensions this week"
- "What are the latest conversations in the Developers community?"
- "Post an announcement to the Business Central community"
- "Search for posts about Business Central errors"
- "Show me my Viva Engage feed"
Important Limitation — Home Network Only
This MCP server works with your home network only (the primary organization network tied to your login credentials).
Why can't it access external networks?
Viva Engage supports two types of networks:
- Home network — your primary organization (e.g. your company's Viva Engage)
- External networks — communities hosted by other organizations you're a guest member of (e.g. Microsoft's BC Partners community)
The Yammer REST API (www.yammer.com/api/v1) only returns data from your home network regardless of what parameters or tokens are used. External networks require browser session cookies that are only available after a full web-based login — something that cannot be replicated through API calls alone.
Microsoft's newer Graph API (graph.microsoft.com) also does not support external networks — it only works with networks in native mode within your own tenant.
This is a Microsoft API limitation, not a limitation of this MCP server. If Microsoft expands their API to support external networks in the future, this server can be updated to support them.
How It Works
You ask Claude a question
↓
Claude calls this MCP server
↓
Server calls Viva Engage API using YOUR login
↓
Returns data only YOU can already see
↓
Claude answers your question
Privacy: Every user registers their own Azure app and uses their own Microsoft credentials. No data ever goes through anyone else's infrastructure. Your login token is stored encrypted on your machine only.
Prerequisites
- Node.js v18 or higher
- Claude Desktop
- A Microsoft work/school account with Viva Engage access
Step 1 — Register Your Own Azure App
This is a one-time setup (~10 minutes). You register your own app under your own Microsoft account. No one else can use your app — it is tied to your credentials only.
- Go to portal.azure.com and sign in with your work/school Microsoft account
- Search for "App registrations" → click "+ New registration"
- Fill in:
- Name:
mcp-viva-engage(or anything you like) - Supported account types:
Multiple Entra ID tenants→ selectAllow all tenants - Redirect URI:
- Platform =
Public client/native (mobile & desktop) - URI =
http://localhost
- Platform =
- Name:
- Click "Register"
- On the overview page, copy and save both of these:
- Application (client) ID
- Directory (tenant) ID
Add API Permissions
- In the left sidebar click "API permissions"
- Click "+ Add a permission" → click the "All APIs" tab → search for "Yammer"
- Click "Yammer" → click "Delegated permissions"
- Check these four permissions:
- ✅
user_impersonation - ✅
Community.Read.All - ✅
EngagementConversation.ReadWrite.All - ✅
Storyline.ReadWrite.All
- ✅
- Click "Add permissions"
Note for work/school accounts: Your organization's IT admin may need to grant consent for this app before you can log in for the first time. This is a one-time step. You can submit an approval request directly from the login screen when you first run the server.
Step 2 — Clone and Build
# Clone the repo
git clone https://github.com/pranav4186/mcp-viva-engage.git
cd mcp-viva-engage
# Install dependencies
npm install
# Build
npm run build
Step 3 — Configure Your Credentials
Copy the example env file:
# Windows
copy .env.example .env
# Mac/Linux
cp .env.example .env
Open .env and fill in your values from Step 1:
AZURE_CLIENT_ID=your-application-client-id-here
AZURE_TENANT_ID=your-directory-tenant-id-here
Important: Never commit your
.envfile to GitHub. It is already in.gitignoreto prevent this.
Step 4 — Connect to Claude Desktop
Open your Claude Desktop config file:
- Windows:
%APPDATA%\Claude\claude_desktop_config.json - Mac:
~/Library/Application Support/Claude/claude_desktop_config.json
Add the viva-engage server to the mcpServers section:
{
"mcpServers": {
"viva-engage-mcp": {
"command": "node",
"args": ["C:/path/to/mcp-viva-engage/dist/server.js"],
"env": {
"AZURE_CLIENT_ID": "your-application-client-id-here",
"AZURE_TENANT_ID": "your-directory-tenant-id-here"
}
}
}
}
Replace
C:/path/to/mcp-viva-engagewith the actual path where you cloned the repo. Windows paths can use forward slashes/or double backslashes\\.
Fully quit and restart Claude Desktop after saving.
Step 5 — First Login
The first time you ask Claude something about Viva Engage:
- A browser window automatically opens with Microsoft's login page
- Sign in with your work/school Microsoft account
- Approve the permissions
- Browser shows "Login successful!"
- Claude completes your request
Your session is kept encrypted on your machine. When you restart Claude Desktop you may need to log in again.
Available Tools
| Tool | Description |
|---|---|
get_networks |
Get all Viva Engage networks you belong to |
get_communities |
List all communities in your home network |
get_community_by_id |
Get details of a specific community by ID |
get_community_messages |
Get posts in a community — accepts name or ID, supports pagination |
get_recent_messages |
Get recent messages across all communities — great for daily summaries |
get_thread |
Get all messages in a conversation thread |
search_messages |
Search posts across your home network — supports date filtering |
post_message |
Post a new message to a community — accepts name or ID |
reply_to_message |
Reply to an existing conversation thread |
get_storyline_feed |
Get your personal Viva Engage storyline feed |
Example Prompts
| What you ask Claude | What happens behind the scenes |
|---|---|
| "Summarize what's new today" | get_recent_messages with hours_ago: 24 |
| "What's new this week in ISV Collaboration?" | get_community_messages with name "ISV Collaboration" |
| "Find AL Extension posts this week" | search_messages with from_date set to 7 days ago |
| "Post to Developers group" | post_message with name "Developers" — no ID needed |
| "Summarize the last 10 posts in Business Central" | get_community_messages with name "Business Central" |
Security & Privacy
| 🔐 Your own Azure app | Every user registers their own app — no shared infrastructure |
| 🔐 Your own credentials | You log in with your own Microsoft account |
| 🔐 Encrypted token storage | Token is encrypted using Windows DPAPI (or plain file on Mac/Linux) |
| 🔐 No secrets in code | Client ID and Tenant ID stay in your local .env file only |
| 🔐 Delegated permissions | Server can only see what you can already see in Viva Engage |
| 🔐 Password never touched | Microsoft handles authentication entirely |
Project Structure
mcp-viva-engage/
├── src/
│ ├── auth/
│ │ └── auth.ts # Microsoft OAuth login with DPAPI encrypted cache
│ ├── yammerClient.ts # Viva Engage / Yammer API client
│ └── server.ts # MCP server and tool definitions
├── .env.example # Template — copy to .env and fill in your values
├── .gitignore # Prevents .env from being committed
├── package.json
├── tsconfig.json
└── README.md
Troubleshooting
"Approval required" on first login
Your organization requires admin consent for new apps. Ask your IT admin to approve the app in Azure portal → Enterprise Applications → find
mcp-viva-engage→ Grant admin consent. This is a one-time action.
Server not showing in Claude Desktop
Fully quit and restart Claude Desktop after editing the config file. Verify the path to
dist/server.jsis correct.
Login required again after restart
The encrypted token may have expired. Simply log in again — the session will be cached after login.
Only seeing home network communities
This is expected — see the Important Limitation section above. The Yammer REST API only supports home network access.
Path issues on Windows
Use either forward slashes
C:/path/to/dist/server.jsor double backslashesC:\\path\\to\\dist\\server.jsin the config file.
get_recent_messages is slow
This tool scans all your communities one by one. If you have many communities it may take 10-20 seconds. This is normal — the Yammer API does not support bulk message fetching.
Known Limitations
- Home network only — External networks (e.g. Microsoft BC Partners community) are not accessible via any public Microsoft API. See explanation above.
- Rate limiting — Yammer API allows 10 requests per user per app per 30 seconds.
get_recent_messagesmay hit this limit if you have many communities. - Read-only search — Search results are limited to what Viva Engage indexes.
- No draft support — Messages posted via
post_messageare published immediately. There is no draft or preview mode in the Yammer API.
Roadmap
- [ ]
get_my_info— get current user profile for better context - [ ]
like_message— like/unlike posts - [ ]
get_community_members— list members of a community - [ ] Convert to a remote connector so users don't need local setup
- [ ] Windows Credential Manager support for even more secure token storage
- [ ] Support external networks if Microsoft adds API support in future
Changelog
v1.1.0
get_community_messagesnow accepts community name or ID (no more looking up IDs)post_messagenow accepts community name or IDsearch_messagesnow supports date filtering (from_dateparameter)- New
get_recent_messagestool — scan all communities for recent activity - Message bodies now return clean plain text (HTML tags stripped)
- Pagination support added to
get_community_messages
v1.0.0
- Initial release
License
MIT — see LICENSE for details.
Disclaimer
This is an independent open source project and is not affiliated with, endorsed by, or supported by Microsoft or Viva Engage. Use at your own risk.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。