MCP server · personal LinkedIn profiles
Your agent already writes your posts. Give it analytics that answer back.
Claude, Cursor, whatever you draft in — right now it is writing from vibes. LinkIntel hands it your real LinkedIn post analytics, read-only, so the next draft is an informed one.
Personal profiles, not Company PagesRead-only, never posts as youCancel in two clicks
✓ linkintel connected · 142 posts measured · read-only
One MCP, your harness
- Claude
- Claude Code
- ChatGPT
- Cursor
- VS Code
- Zed
- Windsurf
- OpenClaw
What you can ask
Questions, not dashboards.
Every question maps to a real tool your agent calls, mid-draft, without leaving the thread. It reads the numbers and answers in a sentence you can act on.
“How did my posts do last month?”
get_performance_summary“Which post got the most reach?”
get_top_posts“Better or worse than last month?”
compare_periods“Which days were unusually good?”
detect_anomalies“What do my best posts have in common?”
get_content_insights“How did that one post actually do?”
get_post_performance“What does my reach look like day by day?”
get_daily_trend“How fresh is this data, really?”
get_data_status
It will also tell you when the data cannot answer the question, which is more often than most analytics tools admit.
Proof
Real numbers, not a demo account.
Everything on this page came out of the founder’s own LinkedIn profile, read through the same API you will connect to. There is no seeded demo org behind it.
- 20,806
- impressions, 30 days
- 1.27%
- engagement, up 63%
- 14
- tools your assistant can call
Setup
Four steps. One evening's coffee.
No dashboard to learn, no tracking script, and nothing to install on LinkedIn itself.
- 01
Subscribe, then connect
One read-only permission, granted on LinkedIn's own consent screen. We never see your password, and LinkIntel can never publish as you.
- 02
Bring your post history
Typefully, Postiz, a webhook from anything else, or LinkedIn's own .xlsx export. LinkedIn will measure a post, but its API cannot list which posts you published.
- 03
Point your assistant at it
One command, or one block of JSON if your client uses a config file. From then on you ask in the assistant you already have open.
- 04
Ask, learn, repeat
Compare periods, rank posts, find the odd days. It stays current on its own, and says so when the data is incomplete.
$ claude mcp add --transport http linkintel https://www.getlinkintel.com/api/mcp \
--header "Authorization: Bearer YOUR_KEY"Claude Desktop, Cursor or another JSON-config client? The equivalent block is in the MCP docs. Want your agent to do the whole thing? Hand it SKILL.md.
Integrations
Your posts come from where you publish them.
LinkedIn will measure any post you name but will not tell anyone which posts you published. So the list comes from elsewhere: the tool you publish through, a webhook from anything else, or LinkedIn's own export.
- API key · keeps working
Typefully
Import every LinkedIn post you published through Typefully, then measure it with LinkedIn's official numbers.
How it works - API key · self-hosting supported
Postiz
Import every LinkedIn post Postiz published, cloud or self-hosted, then measure it with LinkedIn's official numbers.
How it works - Push · any tool
Webhook
Point any tool or script at one endpoint and each post registers as it publishes. Covers everything with no integration of its own.
How it works - Upload · one-off
LinkedIn analytics export
Upload LinkedIn's own 90-day creator analytics export. Works immediately, no third-party account needed.
How it works
Publishing through something not listed? The webhook takes a post URL from any tool that can make an HTTP request, which is very nearly all of them.
The permission
One scope. Read-only. Yours to revoke.
We hold no write permission of any kind. That is a matter of capability, not policy.
It can
- Read impressions, reach and engagement on posts you published
- Read reactions, comments, reshares, saves, sends and link clicks
- Keep that history current so month-on-month comparisons work
It cannot
- Post, comment, react, message or follow as you. Ever.
- See your messages, connections, feed or profile viewers
- See anybody else's posts, or any Company Page
The single scope requested. Remove it on LinkedIn’s permitted-services page any time, without asking us. Your metrics are never sold, never shared, and never used to train models.
r_member_postAnalyticsPricing
One plan. Nothing to forget to cancel.
No trial, because there is nothing to trial your way out of: it either answers your questions in week one, or you cancel in week one.
Member post analytics
$39/month
One personal LinkedIn profile. For people who publish on their own name and want to know what actually happened.
Get startedCancel whenever. It runs to the end of the month you paid for.
LinkedIn’s API allowance caps how many profiles we keep measured daily. If we are full, you will be told before you are asked to pay.
Everything included
- Your post analytics, measured through LinkedIn's official API
- History from Typefully, Postiz, a webhook, or a LinkedIn export
- Summaries, daily trends and period comparisons
- Unusually strong or weak days surfaced, not hunted for
- Posts ranked by reach and by engagement rate
- Asked in plain language, through the client you already use
- A web view too, if you would rather just look at it
Checkout first, then connect LinkedInData never sold or used for trainingDelete it on request
Writing
On reading LinkedIn numbers well
Does LinkIntel access my LinkedIn account?
Yes, in one narrow way. You grant a single read-only permission, r_member_postAnalytics, on LinkedIn's own consent screen. It lets us read performance metrics for posts you published yourself. We never see your password: LinkedIn handles the sign-in and only hands us a token for that one permission.
Can LinkIntel post, comment or react on my behalf?
No. We hold no write permission of any kind, so it is not a matter of policy but of capability: the integration cannot publish, comment, react, message or follow. If you ever see LinkIntel ask for a write permission on LinkedIn's consent screen, something is wrong and you should decline it.
What exactly can it see?
Performance metrics for your own posts: impressions, members reached, reactions, comments, reshares, saves, sends and link clicks. It cannot see your messages, your connections, your feed, your profile viewers, or anybody else's posts.
How do I disconnect it?
Go to LinkedIn's permitted services page and remove LinkIntel. You do not need to ask us or wait for us, and it takes effect immediately. You can also ask us to delete the metrics we have already stored.
What does it cost?
$39 a month, and there is no trial, so there is nothing to remember to cancel. You pay before connecting your LinkedIn account. Cancel whenever you like and it keeps working until the end of the month you have paid for.
Why is there no free tier?
LinkedIn's API allowance limits how many profiles we can keep synced each day, so places are genuinely finite. A free tier would spend one of them on somebody who tried it once, at the expense of somebody who wanted it. If we are full, we will tell you before taking any money.
Do I need Claude to use this?
LinkIntel is delivered as an MCP server, so it is designed to be asked questions through an AI assistant that supports MCP, such as Claude. There is also a web view of the same data if you would rather look at it directly.
Who owns my data?
You do. Your metrics are stored against your own account, visible only to you, and are never sold, shared with third parties, or used to train models. You can request deletion at any time.
Is my data secure?
Metrics are stored in our own infrastructure on encrypted, access-controlled cloud hosting, scoped so that one account cannot read another's. We store the minimum needed to answer questions about your posts, and no LinkedIn password ever reaches us.
How current are the numbers?
Your metrics sync automatically, and LinkIntel will tell you how fresh the data is and how many of your posts are measured rather than quietly presenting a partial picture as a complete one. You can also trigger a fresh pull yourself.
I run a LinkedIn company Page, not a personal profile. Can I use this?
No. LinkIntel reads the posts you publish on your own profile. Organization Page analytics is a different LinkedIn API that this product no longer covers.
Your next post is getting drafted anyway. Let it learn from the last hundred.
Checkout first, then you connect LinkedIn. No trial, so nothing to remember to cancel.