Model Context Protocol

The LinkedIn MCP that writes, not just reads

Work LinkedIn from Claude, Claude Code, Codex, Cursor, Kiro, or Open Claw. Every other Claude–LinkedIn integration stops at reading your inbox. Linxa hands your agent the full loop — see your posts and who engaged, reply to an exact comment, source prospects, score them, invite, and sequence — with your ICP already derived from your website, and every LinkedIn request itemised for you to audit.

Read access is free. Write access is $7/month standalone — and already included in Growth and Outreach.

Get my token Read the setup Browse 41 tools

A token is an API key. This is an agent that already knows your business.

Signup takes one paste: your website URL. Linxa reads it, derives your ICP, target titles, customer pains, and value proposition — then hands all of it to your agent along with the token. Your first prompt works on real, relevant data instead of an empty account.

ICP derived for you

Paste a URL once. Your agent inherits who to target and why, no prompt engineering required.

Full write access

Reply to post comments, score leads, send invites at a safe pace, and build sequences — all with approval.

Every request itemised

Full detail on every LinkedIn call the agent makes. Nothing happens that you cannot audit.

Prefer not to wire it up yourself? The same engine runs behind the AI LinkedIn Agent and Smart Inbox. Same account, same data — MCP is just the door for people who already live in their own tools.

Quickstart

Install Extension

Get the Linxa Chrome Extension and pin it.

Generate Token

Go to Settings > MCP to get your key.

Configure & Run

Add LINXA_API_TOKEN and run a test command.

Core Capabilities

Products & ICP

Create products, generate ICP rules, and keep your agent grounded in what you sell.

Find the right leads

Launch a search from a persona or LinkedIn URL, then work only with scored results.

Control sequences

Build steps, attach a lead-search audience, preview everything, then start or pause.

Handle replies

Prioritize conversations, prepare or edit contextual drafts, approve exact recipients, and verify delivery status.

Setup Guide

Generate API Token

1. Open Linxa MCP Setup.

2. Generate and copy your personal MCP token.

3. Keep it private and use it only in your local MCP client configuration.

Connect your MCP client

Follow the detailed setup, security, and troubleshooting guide for your platform.

Connect your MCP client

Guides are available for Claude Code, Codex, Cursor, Kiro, and OpenClaw.

Read the tool reference

Start with reads, then preview configuration and external actions before your agent runs them.

Open the MCP documentation →

What you can ask Linxa MCP

Ask in natural language. Linxa keeps context and exact IDs across the workflow, prepares changes without sending, and stops for approval before external actions.

Work my post signals

Show who commented on my latest post. Draft a helpful reply using my product context, show it to me, and wait for approval before posting.

Prioritize my inbox

What conversation needs my attention? Rank the results, explain why, and do not send anything.

Prepare grounded replies

Prepare replies to Akbar Jaffer, Rajesh K, and Ilija Stepanovic. Save them as drafts, but do not send.

Edit and send exact drafts

Make Rajesh’s draft shorter. Show me all three final messages, then send only the replies I approve.

Build product context

Create a product from my URL. Generate only the missing ICP, pain points, and value proposition, then show me the result.

Find an audience and sequence it

Find people interested in my service URL. Prepare an invitation, first message, and three follow-ups. Show the exact audience, copy, and delays before starting.

Control a running sequence

Show the sequence status. Pause it, make follow-up two shorter, preview the change, and wait for my approval before resuming.

Available Tools

41 tools cover the full controlled LinkedIn workflow — including your own posts and the people behind every reaction or comment. Publishing a post comment or an exact reply always starts with a preview and waits for your approval.

linxa_get_next_actions

Get the one Needs Attention chat that needs you now, with a direct Smart Inbox link; fall back to ten recent follow-ups only if that folder is empty.

linxa_list_my_posts

List your stored LinkedIn posts with content, media, performance, and sync status.

linxa_get_post_engagement

See the actual people who reacted or commented, what they wrote, and whether you already replied.

linxa_comment_on_post

Preview a top-level comment on one of your posts, then publish it only after explicit approval.

linxa_reply_to_post_comment

Preview an exact reply to one commenter, then publish only after explicit approval.

linxa_prepare_replies

Prepare Smart Inbox-quality drafts for exact conversations without sending, while respecting follow-up dates and no-contact decisions.

linxa_send_drafts

Preview full drafts and recipients, then queue only exact approved conversation IDs with a short-lived approval token.

linxa_list_products

Read the products and ICP context that guide your agent.

linxa_search_leads

Start an asynchronous lead search from a persona or LinkedIn search URL.

linxa_set_autopilot

Set schedule, daily limit and ICP threshold — only after an explicit approval.

linxa_set_sequence_audience

Bind a sequence to explicit owned search sources; a missing audience never means all contacts.

linxa_get_action_status

Check whether a queued action was delivered, cancelled, skipped or failed.

Explore every tool and parameter

Security by Design

Your data stays yours. We prioritize strict security protocols.

No Password Sharing

We never ask for or store your LinkedIn password. Access is handled via a secure, user-controlled session token.

Token-Based Access

Your API token is the only key. You can revoke it instantly from your dashboard if needed.

Session Reliability

The MCP endpoint stays reachable, but real-time message sync requires an active browser session or extension.

Minimal Data Storage

We cache minimal metadata to speed up your agent's queries. Message content is fetched on demand.

Troubleshooting

Token Invalid or Expired

Tokens may expire if you log out. Regenerate a new token in the Linxa Dashboard.

Extension Not Installed

The MCP server relies on your browser extension to proxy requests to LinkedIn securely.

Not Signed Into LinkedIn

Ensure you are logged into LinkedIn in the browser where the extension is installed.

Sync Not Updating

Keep the Linxa tab or extension active to ensure continuous background sync for real-time updates.

Rate Limits

If you hit a rate limit, the tool will pause. Wait a few minutes before retrying to protect your account.

Why Linxa MCP?

Solve common AI development headaches instantly.

The VM Setup HeadacheSkip the VM stress—spin up your chat assistant on your own machine in under 5 minutes. Free, secure, zero infra headaches.
Endless Credentials HellOne environment variable, one CLI install. Everything else “just works.” No more secret-vault gymnastics.
The Feature Creep TrapFocus on what matters. Start with chat-driven prospecting—LinkedIn messages, lead capture—in a consolidated “Lite” package.
Over-Engineering AI PipelinesPre-built agent templates mean you skip the pipeline and jump straight to conversation. Plug-and-play sampling of your own data in seconds.
The Hidden Costs of ScaleRun locally or in burst-mode on free tier cloud. No surprise invoices.
Team Onboarding OverheadOne CLI install, one config file—everyone on the same page instantly.
Siloed Tools & Switch FatigueUnified chat interface connects to all your tools via one plugin. Type once, get context everywhere.
Analytics ParalysisInstant natural-language reports (“Show me last week’s leads”) right in your chat. Zero waiting.
Security Audit NightmaresBuilt-in audit trail and least-privilege defaults—your security team can sleep through the night.
Priced so it is never the reason you say noRead access is free. Write access is $7/month standalone — and included at no extra cost in Growth and Outreach.

How MCP is priced

One value unit, no surprises. If you already pay for Linxa, MCP is already included.

MCP read

Free

Read your inbox, conversations, prospects, and goal progress from any MCP client.

MCP write

$7/mo

Everything above, plus sourcing, scoring, invites, sequences, and drafted replies from your agent.

Growth & Outreach

Included

MCP write access comes with both paid plans at no extra cost, alongside the Agent and Smart Inbox.

Get my token

Paste your URL, confirm your ICP, copy the token. About a minute.

LinkedIn MCP FAQs

How do I connect Claude to LinkedIn?

Connect Claude to LinkedIn through the Linxa MCP server. Sign up, paste your website URL so Linxa derives your ICP, then generate an API token and add the Linxa MCP server to Claude Desktop, Claude Code, or any MCP-capable client. Claude can then read your inbox and prospects, and — with write access — score leads, send invites, and draft messages for your approval.

What can the Linxa MCP server do that a read-only LinkedIn integration cannot?

Most Claude–LinkedIn integrations only read. The Linxa MCP server also writes: it can source prospects from posts and signals, score them against your ICP, send connection invitations at a safe pace, build sequences, and draft replies for approval. Every LinkedIn request it makes is fully itemised so you can audit exactly what ran.

How much does the Linxa MCP server cost?

Read access is free. Full write access is $7 per month as a standalone plan, and it is included at no extra cost in the Growth and Outreach plans — so if you already pay for Linxa, MCP is already yours.

Which AI tools work with the Linxa MCP server?

Any client that speaks the Model Context Protocol: Claude Desktop, Claude Code, Codex, Cursor, Kiro, Antigravity, and Open Claw. Linxa also ships a packaged skill for Claude and Open Claw so the agent knows how to use the tools without you writing the prompts.

Can an AI agent see who engaged with my LinkedIn posts and reply?

Yes. Linxa MCP can list your stored posts, return the people who reacted or commented, and prepare either a new post comment or a reply to one exact commenter. Publishing uses a dry-run preview, explicit approval, and an idempotency key.