# Taplio LinkedIn MCP Server

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](./LICENSE)
[![MCP](https://img.shields.io/badge/protocol-MCP-blue.svg)](https://modelcontextprotocol.io)
[![Remote server](https://img.shields.io/badge/type-remote%20(HTTP)-brightgreen.svg)](#remote-taplio-mcp-server)

The Taplio LinkedIn MCP Server is a [Model Context Protocol (MCP)](https://modelcontextprotocol.io/introduction)
server that connects AI tools directly to your [Taplio](https://taplio.com) account and, through it, to your
LinkedIn presence. It gives AI agents, assistants, and chat clients the ability to draft, schedule, and publish
LinkedIn posts, browse what you have already posted, research inspiration from other creators, read your
analytics, and engage: find posts worth commenting on, draft comments and replies, like posts, and send
connection requests, all through natural language.

Point your MCP host at the Taplio server and you can ask things like:

- "Draft a LinkedIn post about our new MCP launch and schedule it for Tuesday 9am."
- "Show me my last 10 posts and which one got the most impressions."
- "Find 20 high-performing posts about `linkedin ghostwriting` from creators under 50k followers."
- "How are my followers and engagement trending over the last 30 days?"

## Use cases

- Turning raw ideas into publish-ready LinkedIn posts, on your own voice and topics.
- Automating draft creation, scheduling, and publishing from any MCP-capable client.
- Building AI-assisted content workflows and reporting on top of your real LinkedIn data.
- Extracting insights and analytics from your posting history without leaving your assistant.

---

## Contents

- [Remote Taplio MCP Server](#remote-taplio-mcp-server)
- [Prerequisites](#prerequisites)
- [Installation](#installation)
  - [Claude Code](#claude-code)
  - [Claude Desktop / claude.ai connectors](#claude-desktop--claudeai-connectors)
  - [Manus](#manus)
  - [VS Code](#vs-code)
  - [Cursor](#cursor)
  - [Windsurf](#windsurf)
  - [Grok](#grok)
  - [Mistral Le Chat](#mistral-le-chat)
  - [Raycast](#raycast)
  - [Smithery](#smithery)
  - [Generic MCP config](#generic-mcp-config)
- [Authentication](#authentication)
- [Toolsets](#toolsets)
- [Tools](#tools)
- [Tool reference](#tool-reference)
  - [Profile](#profile)
  - [Drafts](#drafts)
  - [Publishing](#publishing)
  - [Posts](#posts)
  - [Analytics](#analytics)
  - [Inspiration](#inspiration)
- [Recommended workflow](#recommended-workflow)
- [Library and lifecycle model](#library-and-lifecycle-model)
- [Limits and safety](#limits-and-safety)
- [Troubleshooting](#troubleshooting)
- [Contributing](#contributing)
- [License](#license)

---

## Remote Taplio MCP Server

Taplio is offered as a hosted, remote MCP server. There is nothing to install or run locally: your MCP host
connects to the Taplio endpoint over HTTP and authenticates with OAuth.

| Property        | Value                                                        |
| --------------- | ------------------------------------------------------------ |
| Transport       | Streamable HTTP (remote)                                     |
| Endpoint        | `https://mcp.taplio.com/`                                    |
| Authentication  | OAuth 2.0 (authorize in the browser, no API key to paste)    |
| Scope           | The authenticated user's own Taplio + LinkedIn account       |
| Server hint     | Call [`get_me`](#get_me) first to load identity and settings |

> Because the server is remote and OAuth-based, you never copy a secret into a config file. Access is tied to
> the Taplio account you approve in the browser, and can be revoked from Taplio at any time.

---

## Prerequisites

1. An active [Taplio](https://taplio.com) account with LinkedIn connected.
2. An MCP-capable host (Claude Code, Claude Desktop, Manus, VS Code with Copilot, Cursor, Windsurf, Grok, Mistral
   Le Chat, Raycast, or any client that speaks MCP over HTTP).
3. A browser available on first connect, to complete the OAuth authorization.

---

## Installation

### Claude Code

```bash
claude mcp add --transport http taplio https://mcp.taplio.com/
```

Then run `/mcp` inside Claude Code and authorize `taplio` in the browser when prompted.

### Claude Desktop / claude.ai connectors

**One click:** open
[this link](https://claude.ai/customize/connectors?modal=add-custom-connector&connectorName=Taplio&connectorUrl=https%3A%2F%2Fmcp.taplio.com%2F)
to get the "Add custom connector" dialog already filled in with the Taplio name and URL, then click **Add**.

Or do it manually: in Claude Desktop or on claude.ai, open **Settings -> Connectors -> Add custom connector**, then
enter:

- **Name:** `Taplio`
- **URL:** `https://mcp.taplio.com/`

Approve the OAuth screen. The Taplio tools then appear in the tool picker.

### Manus

Taplio is a native Manus plugin: one click, no server URL to paste.

1. Open the [Manus plugins directory](https://manus.im/app/plugins#connector_b0b849c6-2cab-4a24-b300-4df5de49a4de).
2. Find **Taplio** and click **Connect**.
3. Sign in to your Taplio account and approve access.

The Taplio tools are then available in your Manus tasks.

### VS Code

Add to your user or workspace MCP configuration (`.vscode/mcp.json`):

```json
{
  "servers": {
    "taplio": {
      "type": "http",
      "url": "https://mcp.taplio.com/"
    }
  }
}
```

### Cursor

In `~/.cursor/mcp.json` (or **Settings -> MCP -> Add**):

```json
{
  "mcpServers": {
    "taplio": {
      "url": "https://mcp.taplio.com/"
    }
  }
}
```

### Windsurf

In `~/.codeium/windsurf/mcp_config.json`:

```json
{
  "mcpServers": {
    "taplio": {
      "serverUrl": "https://mcp.taplio.com/"
    }
  }
}
```

### Grok

On [grok.com](https://grok.com), open **Plugins -> New Connector -> Custom**, then enter:

- **Name:** `Taplio`
- **Server URL:** `https://mcp.taplio.com/`

Click **Add Connector** and approve the Taplio sign-in in the popup window. Mention Taplio in your prompt (for
example "Use Taplio to give me my LinkedIn stats of the day") so Grok picks the connector.

### Mistral Le Chat

In Le Chat, open **Context -> Connectors -> Add connector -> Add a custom connector**, then enter:

- **Connector name:** `Taplio`
- **Server:** `https://mcp.taplio.com/`

Authentication is detected automatically (OAuth 2.1). Click **Add connector** and approve the Taplio sign-in. If
Le Chat answers that no Taplio connector is available, name the tool in your prompt (for example "Use the Taplio
connector's get_me tool").

### Raycast

Install the [Model Context Protocol Registry](https://www.raycast.com/raycast/model-context-protocol-registry)
extension, run **Search Servers**, keep the **Official** registry, search for `taplio`, and choose
**Install Server**. Raycast connects over HTTP and opens the Taplio sign-in.

### Smithery

Open [smithery.ai/servers/taplio/linkedin-mcp](https://smithery.ai/servers/taplio/linkedin-mcp), click
**Add to toolbox**, then **Authenticate** on the Taplio connector in your toolbox.

### Generic MCP config

Any client that supports remote MCP servers over HTTP can use:

```json
{
  "mcpServers": {
    "taplio": {
      "type": "http",
      "url": "https://mcp.taplio.com/"
    }
  }
}
```

---

## Authentication

The Taplio MCP Server uses OAuth 2.0. On first use your client opens a browser window where you log in to Taplio
and approve access. The server then acts on behalf of that single account. No API keys, tokens, or LinkedIn
credentials are stored in your client config. Revoke access at any time from your Taplio account settings.

---

## Toolsets

The tools are grouped into functional toolsets. All of them operate on the authenticated user's own account.

| Toolset                      | Description                                                                    |
| ---------------------------- | ------------------------------------------------------------------------------ |
| **Account and analytics**    | Identity, content preferences, posts, and account and per-post analytics.      |
| **Drafts and media**         | Create, read, list, edit, and delete drafts; upload and attach media.          |
| **Schedule and publish**     | Publish a draft now, schedule it, or unschedule it back to drafts.             |
| **Inspiration**              | Search other creators' LinkedIn posts to model content on what works.          |
| **Engage**                   | Find posts and comments worth engaging with, comment, reply, and like.         |
| **Network**                  | Queue connection requests and follow up on them.                               |

---

## Tools

All 35 tools. Read-only tools never change anything; write tools change your Taplio drafts or act on LinkedIn,
and clients that support tool annotations ask for confirmation before running them. Full input schemas are
returned by the server (`tools/list`); the [tool reference](#tool-reference) below details the core tools.

**Account and analytics**

- `get_me` (read-only): Your identity, content preferences, and today's LinkedIn stats. Call it first.
- `list_posts` (read-only): Your LinkedIn posts, newest first, from Taplio and published directly on LinkedIn.
- `get_post` (read-only): One post by Taplio id or LinkedIn urn.
- `get_analytics_overview` (read-only): Account analytics over time: followers, impressions, engagement.
- `get_post_analytics` (read-only): Per-post impressions, likes, comments, and shares.

**Drafts and media**

- `list_drafts` (read-only): Your unpublished LinkedIn post drafts.
- `get_draft` (read-only): One draft by id.
- `create_draft` (write): Create a draft, optionally with media.
- `update_draft` (write): Rewrite a draft or attach media to it.
- `delete_draft` (write): Permanently delete a draft.
- `list_media` (read-only): Media already uploaded to Taplio.
- `create_media_upload_url` (write): Get a one-time upload URL for an image, video, GIF, or PDF.
- `attach_media_from_url` (write): Attach media from a public URL to a draft.
- `pick_media` (write): Interactive panel to attach media from your library, in clients that support MCP Apps.
- `upload_media` (write): Interactive panel to upload a new file and attach it, in clients that support MCP Apps.

**Schedule and publish**

- `schedule_draft` (write): Schedule a draft to publish automatically at a future time.
- `unschedule` (write): Move a scheduled post back to drafts.
- `publish_draft` (write): Publish a draft now. It goes live and cannot be undone.

**Inspiration**

- `search_inspiration` (read-only): Search other creators' LinkedIn posts for proven formats and angles.

**Engage**

- `find_posts_to_comment` (read-only): Posts worth commenting on, from your posts, a curated feed, or a personalized selection.
- `list_comment_feeds` (read-only): The curated feeds available for engagement.
- `find_comments_to_reply` (read-only): Comments on posts that still deserve a reply.
- `list_comment_drafts` (read-only): Your comment drafts, which are also the queue and history.
- `get_comment_draft` (read-only): One comment draft, to confirm whether it was posted.
- `create_comment_draft` (write): Draft a comment or a reply. Nothing reaches LinkedIn yet.
- `update_comment_draft` (write): Rewrite a comment draft before it is committed.
- `commit_comment_draft` (write): Queue a comment to be posted on LinkedIn under your name.
- `cancel_comment_draft` (write): Cancel a queued comment before it is sent.
- `like_post` (write): Queue a like on a post or a comment.
- `get_post_like` (read-only): Status of a like you queued.
- `cancel_like` (write): Cancel a queued like before it is sent.

**Network**

- `list_connection_requests` (read-only): Every connection request you queued, with its status.
- `send_connection_request` (write): Queue a connection request to someone by LinkedIn username.
- `get_connection_request` (read-only): Status of the latest request for a username.
- `cancel_connection_request` (write): Cancel a queued connection request before it is sent.

---

## Tool reference

### Profile

#### `get_me`

Read this first to orient. Returns the user's identity and content preferences so you can tailor output to them
and recognize their own posts: display name and LinkedIn `@handle` (a user may go by several aliases, so check
both), plus `ai_settings` (industry, role, language, target audience, topics, keywords, description) and today's
LinkedIn stats (followers, connections, profile views).

_No parameters._

---

### Drafts

#### `create_draft`

Create a new LinkedIn post draft (unpublished). Publish it later with `schedule_draft` or `publish_draft`.

| Parameter | Type   | Required | Description                                            |
| --------- | ------ | -------- | ------------------------------------------------------ |
| `content` | string | yes      | The text body of the draft. 1 to 3000 characters.      |

#### `get_draft`

Fetch a single LinkedIn post draft by id.

| Parameter | Type   | Required | Description                          |
| --------- | ------ | -------- | ------------------------------------ |
| `id`      | string | yes      | The id of the draft (from `list_drafts`). |

#### `list_drafts`

List the user's unpublished LinkedIn post drafts.

| Parameter | Type    | Required | Description                                             |
| --------- | ------- | -------- | ------------------------------------------------------- |
| `limit`   | integer | no       | Max drafts to return. Defaults to 25, capped at 100.    |
| `cursor`  | string  | no       | Pagination token from a previous response.              |

#### `update_draft`

Update the content of an existing draft. Only drafts can be edited: published or scheduled posts cannot.

| Parameter | Type   | Required | Description                                                    |
| --------- | ------ | -------- | -------------------------------------------------------------- |
| `id`      | string | yes      | The id of the draft to update (from `list_drafts`).            |
| `content` | string | yes      | The new full text body (replaces existing). 1 to 3000 chars.   |

#### `delete_draft`

Permanently delete a LinkedIn post draft.

| Parameter | Type   | Required | Description                                          |
| --------- | ------ | -------- | ---------------------------------------------------- |
| `id`      | string | yes      | The id of the draft to permanently delete.           |

---

### Publishing

#### `publish_draft`

Publish a draft to LinkedIn immediately. This goes live now and cannot be undone.

| Parameter | Type   | Required | Description                                   |
| --------- | ------ | -------- | --------------------------------------------- |
| `id`      | string | yes      | The id of the draft to publish immediately.   |

#### `schedule_draft`

Schedule a draft to be published automatically at a future time.

| Parameter       | Type   | Required | Description                                                                     |
| --------------- | ------ | -------- | ------------------------------------------------------------------------------- |
| `id`            | string | yes      | The id of the draft to schedule (from `list_drafts`).                           |
| `scheduled_for` | string | yes      | ISO 8601 datetime (e.g. `2026-06-10T14:30:00Z`). At least 2 minutes in the future. |

#### `unschedule`

Unschedule a scheduled post, returning it to drafts so you can edit, reschedule, or delete it.

| Parameter | Type   | Required | Description                                          |
| --------- | ------ | -------- | ---------------------------------------------------- |
| `id`      | string | yes      | The id of the scheduled post (from `list_posts`).    |

---

### Posts

#### `list_posts`

List the user's LinkedIn posts (status `scheduled`, `sending`, or `sent`), most recent first. A draft becomes a
post once it is scheduled or published.

| Parameter | Type    | Required | Description                                                              |
| --------- | ------- | -------- | ----------------------------------------------------------------------- |
| `status`  | enum    | no       | Filter by `scheduled`, `sending`, or `sent`. Omit for all three.        |
| `from`    | string  | no       | Only posts on or after this date (YYYY-MM-DD or ISO 8601, UTC).         |
| `to`      | string  | no       | Only posts on or before this date (YYYY-MM-DD or ISO 8601, UTC).        |
| `limit`   | integer | no       | Max posts to return. Defaults to 25, capped at 100.                      |
| `cursor`  | string  | no       | Pagination token from a previous response.                              |

#### `get_post`

Fetch a single LinkedIn post by id. A post has status `scheduled`, `sending`, or `sent`; drafts are separate
(use `get_draft`).

| Parameter | Type   | Required | Description                                |
| --------- | ------ | -------- | ------------------------------------------ |
| `id`      | string | yes      | The id of the post (from `list_posts`).    |

---

### Analytics

#### `get_analytics_overview`

Get the user's LinkedIn analytics overview (followers, impressions, engagement). Defaults to today when no date
range is given.

| Parameter     | Type   | Required | Description                                                                                                    |
| ------------- | ------ | -------- | -------------------------------------------------------------------------------------------------------------- |
| `from`        | string | no       | Start date (YYYY-MM-DD, UTC). Defaults to today; range capped at 90 days.                                       |
| `to`          | string | no       | End date (YYYY-MM-DD, UTC). Defaults to today; range capped at 90 days.                                         |
| `metrics`     | string | no       | Comma-separated subset: `followers, connections, profile_views, impressions, likes, comments, shares`. Omit for all. |
| `granularity` | enum   | no       | Time-bucket size; only `day` is supported (default).                                                           |

#### `get_post_analytics`

List the user's per-post analytics (impressions, likes, comments, shares), defaulting to the last 7 days.

| Parameter | Type    | Required | Description                                                            |
| --------- | ------- | -------- | --------------------------------------------------------------------- |
| `from`    | string  | no       | Start date (YYYY-MM-DD, UTC). Defaults to 7 days ago; range capped at 90 days. |
| `to`      | string  | no       | End date (YYYY-MM-DD, UTC). Defaults to today; range capped at 90 days. |
| `limit`   | integer | no       | Max posts to return. Defaults to 25, capped at 100.                    |
| `cursor`  | string  | no       | Pagination token from a previous response.                            |

---

### Inspiration

#### `search_inspiration`

Search LinkedIn posts from other creators to use as inspiration.

| Parameter        | Type    | Required | Description                                                                       |
| ---------------- | ------- | -------- | --------------------------------------------------------------------------------- |
| `query`          | string  | no       | Free-text search over post content. Omit to browse top posts.                     |
| `lang`           | string  | no       | Filter by ISO 639-1 language code (e.g. `en`, `fr`).                               |
| `min_likes`      | integer | no       | Only posts with at least this many likes. 0-1000; defaults to 50.                 |
| `min_comments`   | integer | no       | Only posts with at least this many comments. 0-1000; defaults to 0.               |
| `min_char_count` | integer | no       | Only posts with at least this many characters. 0-10000; defaults to 500.          |
| `max_followers`  | integer | no       | Only posts whose author has at most this many followers. Defaults to 100000.      |
| `min_days_old`   | integer | no       | Only posts at least N days old (0-365); defaults to 1. Overrides `from`/`to`.      |
| `max_days_old`   | integer | no       | Only posts from the last N days (0-365). Overrides `from`/`to`.                    |
| `from`           | string  | no       | Only posts on or after this date. Ignored if `min`/`max_days_old` is set.          |
| `to`             | string  | no       | Only posts on or before this date. Ignored if `min`/`max_days_old` is set.         |
| `limit`          | integer | no       | Max posts to return. Minimum 1; defaults to 30.                                    |

---

## Recommended workflow

1. Call [`get_me`](#get_me) once at the start of a session to load the user's voice, topics, language, and stats.
2. Use [`search_inspiration`](#search_inspiration) to ground new content in what is currently working in the niche.
3. Draft with [`create_draft`](#create_draft), iterate with [`update_draft`](#update_draft).
4. Ship with [`schedule_draft`](#schedule_draft) (preferred) or [`publish_draft`](#publish_draft) (immediate).
5. Report with [`get_analytics_overview`](#get_analytics_overview) and
   [`get_post_analytics`](#get_post_analytics), and cross-reference with [`list_posts`](#list_posts).
6. Engage with `find_posts_to_comment` and `find_comments_to_reply`, draft with `create_comment_draft`, and only
   call `commit_comment_draft`, `like_post`, or `send_connection_request` once the user agrees.

---

## Library and lifecycle model

A **draft** and a **post** are the same item at different stages of its life:

```
create_draft ─┐
              ├─> [ DRAFT ] ──schedule_draft──> [ scheduled ] ──(auto)──> [ sending ] ──> [ sent ]
update_draft ─┘        │                              │
                       └──────publish_draft───────────┴──(immediate)──> [ sending ] ──> [ sent ]

                       [ scheduled ] ──unschedule──> [ DRAFT ]
```

- Drafts live in the **drafts** list (`list_drafts`) and are the only editable stage.
- Scheduling or publishing a draft turns it into a **post** and moves it to the **posts** list (`list_posts`).
- `unschedule` moves a scheduled post back to drafts; a `sent` post cannot be un-sent.

---

## Limits and safety

- **Post length:** 1 to 3000 characters (`create_draft`, `update_draft`).
- **Scheduling:** `scheduled_for` must be at least 2 minutes in the future, ISO 8601, UTC recommended.
- **Analytics range:** capped at 90 days per query; paginate with `cursor` for large result sets.
- **Irreversible actions:** `publish_draft` goes live immediately and `delete_draft` is permanent. Confirm intent
  before calling either.
- **Queued engagement:** `commit_comment_draft`, `like_post`, and `send_connection_request` queue the action; it
  reaches LinkedIn minutes or hours later, paced to keep the account safe. Each can be cancelled until it is sent
  (`cancel_comment_draft`, `cancel_like`, `cancel_connection_request`), and is irreversible once it is.
- **Scope:** every tool acts only on the authenticated user's own Taplio and LinkedIn account.

---

## Troubleshooting

- **Tools do not appear:** confirm the server URL is `https://mcp.taplio.com/` and that you completed the OAuth
  browser step. Re-run `/mcp` (Claude Code) or reconnect the connector.
- **`401` / auth errors:** your session may have expired or been revoked. Reconnect and re-authorize in the
  browser.
- **Cannot edit a post:** only drafts are editable. If it is already scheduled, call `unschedule` first; if it is
  `sent`, it can no longer be changed.
- **Empty analytics:** newly published posts take time to accumulate metrics; also check your date range against
  the 90-day cap.

---

## Contributing

This repository documents the Taplio LinkedIn MCP Server. See [CONTRIBUTING.md](./CONTRIBUTING.md) for how to
report issues with the documentation or suggest improvements. For product support, contact Taplio.

---

## License

This project is licensed under the terms of the MIT License. See [LICENSE](./LICENSE).