MCP connector

Connect your Linnx account to the AI assistant of your choice, over the Model Context Protocol.

Endpoint

URLhttps://demo.linnx.ai/api/mcp

TransportStreamable HTTP

AuthOAuth 2.1, authorization code with PKCE (S256)

RegistrationDynamic Client Registration (RFC 7591)

Use this URL to connect Linnx to the AI client of your choice. You then sign in to Linnx to authorise the connection, and your Linnx account data is connected.

Once connected, the client you authorised can read the Linnx data your account has access to. How that client stores and uses it is covered by their terms, not ours. You can revoke access at any time from the client you connected.

How to connect

In your assistant, add a custom MCP connector.
Paste the URL above.
It sends you to Linnx to sign in. Approve the connection.

Access is per person. Two people in the same workspace each connect separately, and each sees only their own Linnx account.

Guides for common AI clients

Menu paths move between releases, so here are links to the client’s own instructions rather than ours.

Claude, custom connectorshttps://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp

Claude, when your organisation controls connectorshttps://support.claude.com/en/articles/15537633-authorize-mcp-connectors-for-your-entire-organization

ChatGPT, when your organisation controls connectorshttps://help.openai.com/en/articles/12584461-developer-mode-and-mcp-apps-in-chatgpt

Cursorhttps://cursor.com/docs/mcp

VS Codehttps://code.visualstudio.com/docs/agent-customization/mcp-servers

VS Code, when your organisation controls connectorshttps://code.visualstudio.com/docs/enterprise/ai-settings

If you are on a work account

Whether you can add a connector at all is a setting inside your AI client, not inside Linnx. Most clients let whoever administers the account there decide which MCP connectors its people may use, so if Linnx does not appear after you connect, it has probably been restricted on that side.

There is nothing for us to switch on: Linnx has no admin console, no way to enable itself across an organisation, and no setting we can change on your behalf. Below is where each client keeps the control, so you know who to ask.

Claude: which connectors are available is decided by whoever owns your Claude organisation, in Claude’s own settings. If Linnx is not showing up, ask them.
ChatGPT: custom MCP connectors are a feature of ChatGPT’s own business, enterprise and edu plans, and only whoever administers your ChatGPT account can turn one on, in ChatGPT’s own settings.
Cursor: which MCP servers are allowed is set in Cursor’s own Team Settings, under MCP Configuration, by whoever administers your Cursor account.
VS Code: with Copilot Business or Enterprise, MCP is off until whoever administers your Copilot account turns it on, in GitHub’s own Copilot settings.

You do not need the MCP to use Linnx. Everything the MCP can do, you can also do yourself in the app.

What the MCP can do

Connecting Linnx to an AI client over the MCP means that client can access your Linnx data. All AI processing happens in that client, not in Linnx.

Categories are your own. A chat can carry up to three. A post and a connection request carry one each.

Reads are side-effect-free. Writes only ever change your Linnx data: they are reversible, idempotent, and none of them contacts LinkedIn.

Read (15)

list_chatsYour chats, newest first. Filter by category, spam or archived state, or search text.

get_chatOne chat with its full message history.

search_messagesSearch the text of your messages across a date range.

inbox_countsHow many chats sit in each state and in each category.

list_categoriesYour categories for chats, posts or connection requests. One kind at a time.

list_connectionsYour connections, with each person's name, headline and LinkedIn profile link, plus their chats and groups.

list_groupsYour connection groups.

list_requestsYour received and sent connection requests. Read only.

search_postsSearch the text of your posts across a date range.

list_comments_givenThe comments you have left on posts, with the text of each and who the post was by. Includes replies on your own posts.

post_metricsImpression, reaction and comment totals over a period, and a heatmap of when you post.

connection_analyticsRequest volume in and out, acceptance rates, and net new connections.

message_analyticsReply rates, response times, and who is waiting on whom.

follower_growthYour follower count over time.

dashboard_statsThe headline numbers from your dashboard.

Write (14)

save_draftWrite a reply into the composer for you to review and send. An empty string clears it.

set_chat_categoriesSet the categories on a chat, replacing whatever it had, and leave a note.

set_chat_stateMark a chat as spam, unspam it, archive it or unarchive it.

create_categoryCreate a category for chats, posts or connection requests.

update_categoryRename a category or change its colour.

delete_categoryDelete a category. Anything carrying it keeps the item and loses the tag.

set_post_categoryPut a post into a category, or clear it with null. One per post.

set_request_categoryPut a connection request into a category, or clear it with null. One per request.

create_groupCreate a group from your connections. Creating one whose name already exists returns the existing group.

delete_groupDelete a group. Connections in it are not deleted, just no longer grouped.

add_to_groupAdd connections to a group. This also favourites them.

remove_from_groupRemove connections from a group.

set_favouriteMark connections as a favourite.

remove_favouriteRemove connections from favourites.

Categories are the user's own

There are three separate vocabularies: CHAT for inbox tags, POST for post categories, REQUEST for connection requests. Linnx ships none of them: every category in an account was created by the user, in the app or through this connector, and they can rename, delete and add to them freely.

So call list_categories before tagging and use the slugs it returns. Do not assume a fixed list, and do not invent taxonomy: create a category only when the user asks for one. An unknown slug is dropped rather than guessed at, and the response names it in unknownCategories so you can create it and retry.

Deleting a category does not delete the things tagged with it, but it does clear the tag from them: every chat, post and connection request that carried it has the slug removed, and the delete response reports how many were cleared.

What it cannot do

Rank, score or judge anythingLinnx does not classify, rank or judge your content or your contacts. Your categories are your own, and all AI processing happens in the client you connected, never in Linnx.

Send a LinkedIn messageIt can write a draft into the composer. You press send, in Linnx.

Accept, decline or withdraw a connection requestYou do that yourself, in Linnx.

Reach anyone else's Linnx accountOnly the account you authorised, and only while you allow it.

Actions performed via the MCP only affect your data in Linnx. Anything that reaches LinkedIn remains a human action, taken by you from within Linnx.

Privacy

The connector reads your Linnx data and returns it to the client you connected.

If you authorise an AI client to access that data via the MCP, how the client stores, processes and uses it is governed by its own terms and privacy policy, not ours. You are responsible for reviewing those terms and for meeting your own privacy and data protection obligations before you connect. See the privacy policy.

You can revoke MCP access at any time from the AI client you connected it to.

Linnx is intended for business and professional use, as set out in our terms.