For AI Agents

Give your AI agent a LinkedIn brain

AI on both sides of your LinkedIn workflow: Lina, the assistant built in — and MCP/REST tools so Claude, ChatGPT, Cursor, Codex and OpenClaw can search your saves and draft, schedule, publish and analyze content.

LinkedMash is an MCP server and REST API for LinkedIn saved posts. AI agents such as Claude, ChatGPT, Cursor, and Codex connect through one MCP config block or bearer-token REST calls to api.linkedmash.com/v1, then search your saved posts, draft and schedule content, publish to LinkedIn, and read analytics.

Last updated: July 9, 2026

Read the agent docs

Open in ChatGPT Open in Claude

CYour agentMCP

Linkedmash connected

Find my best saved posts on cold outreach and schedule a draft for 9am.

linkedmashsearch_bookmarks6 results

linkedmashschedule_postMon 9:00

Drafted from your 6 strongest posts and scheduled it for 9am Monday. Want me to draft this week's follow-up too?

Capabilities

Everything your agent can do

Each capability maps to an MCP tool and a REST endpoint, so your agent can run your whole LinkedIn workflow end to end.

Saved posts

Search and organize your saved LinkedIn posts — your agent reads your whole library.

search_bookmarks · GET /v1/bookmarks

Content

Draft and create LinkedIn posts from your best saves, in your own voice.

create_draft · POST /v1/drafts

Schedule & publish

Queue posts to the right slots and publish straight to LinkedIn.

schedule_post · POST /v1/schedule

Analytics

Pull impressions, engagement and follower growth to see what lands.

get_analytics · GET /v1/analytics

Growth profile

Read the positioning, pillars and audience insights Lina builds — so your agent writes with her context.

get_growth_digest · GET /v1/growth/digest

Meet Lina

The AI assistant already inside

External agents plug in over MCP — but LinkedMash ships with its own: Lina, a LinkedIn growth assistant that knows your library and sounds like you.

Content drafts

Drafts written in your voice, from your saves

Ask Lina for a post and she doesn't start from a blank page. She pulls the strongest material from your saved posts, writes it in your voice — your tone, your formats — and hands you a draft you can schedule or refine in the Composer.

Grounded in your saved posts, with citations
Your tone and formats — not generic AI copy
One tap to schedule or open in the Composer

Linain-app Voice: yours

Draft a post from my saves about employee-led content — one experiment I can run this week.

We tested employee-led posts for 2 weeks. No brand account came close.
Here's the experiment, the numbers, and what I'd change…

◆ Grounded in 4 saved posts · written in your voice

Schedule 9:00Edit in Composer

Learns you

She studies your posts, not just your prompts

In a two-minute setup, Lina imports your own LinkedIn posts to learn how you write and what your audience responds to. From that she builds your positioning, your content pillars, and what to do next — and keeps refining it as you chat.

Voice learned from your published posts
Positioning, pillars & audience insights
Proactive next steps — never a blank home screen

Learned from 48 of your posts + your saves

Your content pillars

AI agents in productionFounder lessonsBuild in public

What lands with your audience

Posts with real numbers 3.2× engagementPersonal experiments 2.4×

Proactive

She shows up before you ask

Lina's home is never blank. A deterministic what's-next engine reads your goal — grow your presence, get more from your saves, or both — and surfaces the moves that matter: unread gems worth your time, a follow-up to a winning post, an empty queue filled from this week's saves.

Goal-aware: curators get reading moves, creators get posting moves
Follow up on winners — she reads your post performance
Cards learn from your clicks and retire what you ignore

GoalGrow my presenceMore from my savesBoth

Learn3 unread saves worth your time this weekLearnYour saves disagree on cold outreach — the contrarian readDoTurn Tuesday’s save into a this-week experimentShareMonday’s post is winning — draft the follow-up

A 2-minute setup, not a form

1Pick your goal
2Import your own posts
3Review positioning & first moves

Grounded on any post

Ask Lina about one post, a selection, or a whole search — answers cite the exact saves they came from.

Findings become drafts

Tell Lina what you learned in the real world; she logs it as a draft — your next post is documentation, not invention.

Your external agents tap the same brain: the growth profile Lina builds is available over MCP and the REST API, so Claude, Cursor or your own automation writes with the same context she has.

Connect

Works with your agent

LinkedMash speaks MCP, so it drops into the agents and editors you already use.

CClaude

⌘Claude Code

▦Cursor

{ }VS Code

∑Codex

◇OpenClaw

◉ChatGPT

Add LinkedMash as an MCP server and your agent gets the tools above. Paste the config, set your token, and you are connected.

Connect your agent

mcp.json

{
  "mcpServers": {
    "linkedmash": {
      "url": "https://mcp.linkedmash.com/api/mcp?token=lm_your_token_here"
    }
  }
}

Skills

Install the LinkedMash skill

Install the LinkedMash skill in one command — it brings the tools, the auth and the know-how to work your library, ready for Claude Code, Cursor and any Agent Skills client.

View the skill

linkedmash/agent-skills

$npx skills add linkedmash/agent-skills

# or, in Claude Code

›/plugin marketplace add linkedmash/agent-skills

Two ways in

MCP for agents, REST for code

Use the path that fits — a conversational agent over MCP, or your own tools over a clean REST API.

MCP

For AI agents

The fastest way to give a chat agent your LinkedIn library and content tools.

Works with Claude, ChatGPT, Cursor and more
Search, draft, schedule and analyze from chat
One config block to connect

MCP docs

REST API

For your code

Build on your saves and content with predictable JSON endpoints and webhooks.

Bearer-token auth at api.linkedmash.com/v1
Search, labels, drafts, schedule and analytics
Webhooks for save, label and sync events

API docs

FAQ

Frequently asked questions

How do I connect Claude to my LinkedIn saved posts?

Add the LinkedMash MCP server to Claude with one config block that points at mcp.linkedmash.com/api/mcp with your API token. Claude can then search your LinkedIn saved posts, draft and schedule content, and read analytics. Setup takes about two minutes.

Is there an MCP server for LinkedIn saved posts?

Yes. LinkedMash runs a hosted LinkedIn saved posts MCP server at mcp.linkedmash.com. It works with Claude, Claude Code, ChatGPT, Cursor, VS Code, and any MCP client, and it needs no local install.

What can an AI agent do with LinkedMash?

An agent can call tools such as search_bookmarks, create_draft, schedule_post, get_analytics, and get_growth_digest. That covers finding saved posts, drafting and scheduling LinkedIn content, publishing, and reading performance data.

Does LinkedMash work with ChatGPT, Cursor, and Codex?

Yes. Any MCP client can connect: ChatGPT, Cursor, VS Code, Windsurf, Codex, and OpenClaw all use the same hosted server URL. Tools that prefer plain HTTP can call the LinkedMash saved posts API at api.linkedmash.com/v1 instead.

How do I install the LinkedMash skill in Claude Code?

Run /plugin marketplace add linkedmash/agent-skills inside Claude Code, or npx skills add linkedmash/agent-skills in a terminal. The skill teaches an agent how to drive the LinkedMash saved posts API without extra prompting.

Can an AI agent export my LinkedIn saved posts to Notion or Google Sheets?

Yes. LinkedMash auto-syncs saved posts to Notion, Google Sheets, Airtable, and Miro, and an agent can trigger CSV, JSON, or PDF exports through the API. Your library stays portable wherever you work.

Get started

Give your agent its LinkedIn brain

Get your API token Read the docs

Or start inside — try Lina free with 20 messages on every account.