A Monday LinkedIn briefing with Microsoft Copilot Studio

The short answer

You can have a Microsoft Copilot agent prepare a Monday-morning LinkedIn briefing: who messaged you, which invitations are waiting, what happened in your notifications, and which replies to draft first. It runs in Copilot Studio, connected to Finchreach, a remote MCP server for LinkedIn (MCP is a standard plug that lets AI apps use outside tools).

Two honest limits. First, the consumer Microsoft Copilot app does not support custom MCP servers, so this needs Copilot Studio with a work account and a Copilot Studio license or trial. Second, Finchreach connects one personal LinkedIn account per key, so build this agent for yourself, not as a shared team bot. Any reply the agent drafts waits in your approval queue before it is sent.

Can Microsoft Copilot connect to LinkedIn through MCP?

As of October 2026:

• Consumer Microsoft Copilot (copilot.microsoft.com and the Windows app): no custom MCP servers. Microsoft has not documented a way to add one.
• Copilot Studio: yes. It supports MCP servers over Streamable HTTP with no auth, an API key (in a header or query string) or OAuth 2.0. This post uses the API key route. Source: Microsoft Learn.
• GitHub Copilot in VS Code is a different product and connects too; see the GitHub Copilot post.

Your organization’s Power Platform data policies also apply in Copilot Studio. If adding a custom MCP tool is blocked, that is an admin decision to discuss with IT.

Why a weekly LinkedIn briefing?

For account executives, partner managers and consultants, LinkedIn is a second inbox that rarely gets the same attention as email. Messages from prospects sit next to invitations and notification noise. A short Monday briefing turns that into a list of decisions:

• Conversations that need a reply, with a one-line summary each.
• New connections from your notifications, grouped into “relevant to my accounts” and “other”.
• Notifications worth acting on, such as comments on your posts.
• Suggested first replies, drafted and ready to approve.

Reading is all the briefing needs. Writing is optional, and every write goes through approval.

How does the Copilot Studio agent build the briefing?

1. Reads recent conversations with linkedin_get_conversations and opens the important ones with linkedin_get_messages.
2. Reads linkedin_get_notifications for comments, mentions, accepted invitations and other activity.
3. Writes the briefing in chat, using the instructions you gave the agent: your accounts, your territory, what counts as urgent.
4. If you ask, drafts replies and queues them with linkedin_send_dm. Each returns pending_approval until you approve it in Finchreach.

Finchreach has no scheduler and Copilot Studio triggers are outside the scope of this post, so the simplest version is to open the agent on Monday and ask for the briefing.

What instructions and prompts should the agent use?

Put the stable part in the agent’s instructions, for example: “You prepare a LinkedIn briefing for an account executive covering mid-market retail in the Northeast US. Treat messages that mention pricing, a trial, procurement or a meeting as urgent. Never send anything without showing it to me first.”

Ask your assistant“Give me my LinkedIn briefing for the past 7 days: urgent conversations first, then other replies needed, then new connections grouped by relevance, then notable notifications. One line each.”

Ask your assistant“Draft replies to the three urgent conversations, under 80 words each, and show them to me. Queue only the ones I say yes to.”

Ask your assistant“Draft a short welcome DM to each new connection at a company on my account list and give me the action ids to approve.”

How do I add Finchreach to Copilot Studio?

1. Create a Finchreach account, connect your LinkedIn account on the Accounts page and create an API key.
2. In Copilot Studio, open your agent and go to Tools → Add a tool → New tool → Model Context Protocol.
3. Enter a name and description, and the server URL https://mcp.finchreach.com/mcp.
4. For Authentication choose API key, type Header, header name Authorization. Then Create.
5. Add the tool, create a new connection, and enter Bearer lk_... as the key. Microsoft’s documentation does not say whether Studio adds the Bearer prefix itself, so include it and test with a read-only prompt.
6. Add the tool to the agent and try “show my pending LinkedIn invitations”.
7. See the Copilot guide, the integration page and the docs.

The connection holds your personal API key. Do not publish the agent to colleagues: everyone using it would be reading and acting on your LinkedIn account.

What are the risks?

Automating a personal LinkedIn account carries risk under LinkedIn’s User Agreement. Finchreach’s approval queue and a human pace lower that risk; they do not remove it.

A read-mostly briefing is the gentlest way to start. Keep writes to the handful you would have sent by hand anyway.

There is a data question as well: your LinkedIn messages pass through Copilot Studio and your tenant’s policies. Check that this fits your company’s rules before connecting a work-related LinkedIn inbox.

Frequently asked questions

Can I use this in Copilot in Teams or Outlook? Only to the extent your Copilot Studio agent is available there; we have not verified every channel.

Can the agent post on LinkedIn? Yes, linkedin_create_post publishes a text post after approval, but a briefing agent rarely needs it.

Is there a no-license option? Use a header-capable MCP client instead, such as Claude Code or GitHub Copilot in VS Code. Sales users may like inbox triage.

What is MCP, again? See what a LinkedIn MCP server is.