LinkedIn Monitoring MCP Server

The LinkedIn MCP server thatdoesn't need your LinkedIn account

Octolens monitors LinkedIn posts and comments for your keywords, filters the noise with AI, and serves clean mentions to Claude, ChatGPT, and Cursor over MCP. No session cookies, no scraping on your login, no account risk.

5-minute setup • OAuth sign-in, read-only • MCP on every plan
Your AI assistant · octolens connected
Any high-intent LinkedIn mentions this week we haven't replied to?
octolens · list_relevant_mentionssource: linkedin · 2 results
Priya Nair · VP Marketingbuy_intent

Team is evaluating social listening tools this month. Brandwatch feels like overkill — what are people actually using?

Post · 31 comments · 2h ago

Devon Alvarez · DevRel Leadtestimonial

Comment: “We moved to a tool that catches the LinkedIn posts where nobody tags us — game changer”

Comment · under a post with 213 reactions

Two threads worth acting on. Priya's evaluation post is active and has no vendor replies yet — want me to draft a comment in your voice?

Trusted by brands people talk about online

Cursor logo
Sentry logo
Vercel logo
PostHog logo
Railway logo
Render logo
Supabase logo
PlanetScale logo
Prisma logo
Juicebox logo
Modal logo
Metabase logo
Tally logo
Lovable logo
Medusa logo
CodeRabbit logo
The problem

LinkedIn Is the Hardest Platform to Monitor

It's where B2B buyers ask for recommendations and compare vendors in public — and it's the one platform every monitoring approach struggles with.

01

There is no public API

LinkedIn doesn't offer a keyword-search API to third parties. There's nothing official to build on — which is why almost every "LinkedIn MCP server" you'll find is a workaround.

02

Cookie scrapers risk your account

Most open-source LinkedIn MCPs authenticate with your li_at session cookie. Sessions expire, coverage is limited to what your account can see, and automated access violates LinkedIn's User Agreement — accounts get restricted for it.

03

You're only notified when tagged

Most brand mentions on LinkedIn happen in comment threads under other people's posts. No tag, no notification — the highest-intent conversations are invisible from your own feed.

Octolens fixes this upstream: LinkedIn collection runs on our side, against posts and comments, for the keywords you choose. Your AI tools get clean, scored data over MCP — and your LinkedIn account is never involved.

How it works

From LinkedIn Thread to Answer in Your AI Tool

One pipeline: Octolens does the collection and filtering, MCP does the delivery. You just ask.

1

Add keywords

Your brand, competitors, and the category phrases your buyers use — plus negative keywords for noise.

octolens.com → Keywords
2

Octolens collects

Posts and comments matching your keywords, with a 7-day backfill per keyword and continuous collection after. Hourly refresh on Pro, real-time on Scale.

posts + comments
3

AI scores everything

Every mention arrives scored for relevance and sentiment, tagged with labels like buy_intent, competitor_mention, or complaint.

"relevanceScore": "high"
4

Your assistant queries it

23 MCP tools: list and search mentions, run analytics, manage keywords and feeds — from Claude, ChatGPT, or Cursor.

list_relevant_mentions

Full tool reference in the MCP docs. The same data is available over the REST API and webhooks — on every plan. Prefer alerts and a feed over an AI assistant? See LinkedIn monitoring.

Setup

Connected in Five Minutes

No LinkedIn credentials at any step. You need an Octolens account (7-day free trial) and an MCP-compatible tool.

1

Add your keywords in Octolens

Brand, competitors, category terms. LinkedIn collection (posts + comments) starts with a 7-day backfill.

2

Connect your client

One URL for every tool. Authentication is a browser sign-in to your Octolens account — there's no API key or token to paste.

3

Ask

Try: “show linkedin mentions from today” — you should see your first scored mentions come back.

Anything not covered here — scopes, all 23 tools, troubleshooting — is in the MCP v2 docs. Setup snippets are also in the app under Settings → MCP.
claude mcp add --transport http octolens \
  "https://app.octolens.com/api/mcp/v2"

# then, inside Claude Code:
/mcp   # → octolens → Authenticate

Run the command, then run /mcp inside Claude Code, select octolens → Authenticate, and sign in when the browser opens.

Verify it works

@octolens show linkedin mentions from today

New keywords can take up to 24 hours to index fully — check mentions in the Octolens dashboard first if nothing comes back.

Prompts to steal

What You'll Actually Ask It

Copy-paste these into Claude, ChatGPT, or Cursor once you're connected.

Morning digestShow LinkedIn mentions from the last 24 hours, sorted by engagement

Start the day knowing what's happening on LinkedIn without opening the feed — and losing 30 minutes to it.

Buying intentFind posts and comments where people ask for tool recommendations in our category

“Anyone recommend…” posts are open RFPs with a name, title, and company attached. Being the first helpful reply regularly turns into pipeline.

Invisible mentionsShow comment mentions of our brand under other people's posts this week

The mentions LinkedIn never notifies you about — untagged comments in other people's threads. This is the coverage manual monitoring can't do.

Share of voiceCompare LinkedIn mentions of us vs [competitor] this month, with sentiment

Know your share of the conversation, which threads compare you head-to-head, and which ones have no vendor reply yet.

TriageShow negative LinkedIn mentions from the last 48 hours

Negative posts sit in front of your exact buyer audience. A fast, non-defensive reply usually flips the thread.

Reply draftingDraft a reply to this post in our voice: [paste post]

Draft in your assistant, post it yourself — MCP access is read-only, so a human always stays in the loop.

Try these on your own mentions

7-day free trial · No credit card required

Honest comparison

Your Three Options for LinkedIn via MCP

LinkedIn has no public keyword API, so every option is a trade-off. Here's the honest version.

 Octolens MCPCookie-based LinkedIn MCPManual monitoring
LinkedIn login requiredNoYes — your session cookieYes
Account restriction riskNoneReal riskNone
Keyword monitoring across LinkedInYesLimited & fragileTagged mentions only
Comment coveragePosts + commentsPartialOnly if tagged
AI relevance & sentiment scoringBuilt-inYou build itYour eyeballs
Historical data & trendsYesLimitedNo
Other platforms via the same server15+ sourcesLinkedIn only—
Cost$199/mo ($159/mo annual)Free + your time + account riskFree + hours every week

Cookie-based servers are fine for experimenting with MCP itself. Manual monitoring works until the first untagged comment thread recommends your competitor. If LinkedIn actually matters to your pipeline, monitor it properly.

FAQ

LinkedIn MCP Server & Monitoring, Answered

Setup, account safety, and coverage — everything about monitoring LinkedIn from Claude, ChatGPT, or Cursor.

No. LinkedIn doesn't offer a public API for keyword monitoring, and Octolens never asks for your LinkedIn login, session cookie, or API credentials. Octolens handles LinkedIn data collection on its side — you configure keywords and connect the MCP server with a browser sign-in to your Octolens account.

Both posts and comments. Comment coverage matters more than most people expect: the majority of brand mentions on LinkedIn happen in comment threads under other people's posts, where LinkedIn never notifies you unless you're tagged.

No. Nothing runs through your LinkedIn account — Octolens doesn't use your login or session cookies. This is the key difference from most open-source LinkedIn MCP servers, which authenticate with your personal li_at session cookie and can get your account restricted for automated access.

No — MCP access is read-only by design. You can monitor mentions, analyze them, and draft replies in Claude, but you post them yourself on LinkedIn. That keeps a human in the loop and your account safe from automation flags.

Depends on your plan: hourly refresh on Pro ($199/mo, or $159/mo billed annually), real-time on Scale ($599/mo, or $499/mo billed annually). For LinkedIn, hourly is enough to reply while a post is still in the algorithm's distribution window.

Yes. Add competitor names and category phrases like "[competitor] alternative" as keywords. You'll see complaint threads, comparison discussions, and evaluation posts — often before the competitor does.

Claude Code, Claude Desktop and Web, ChatGPT, Cursor, and any other MCP-compatible client. It's one remote server URL with OAuth sign-in — there's no API key to manage and nothing to run locally.

Every new keyword gets a 7-day backfill, then collects continuously — your archive builds from the day you add a keyword. That's enough history for trends, sentiment shifts, and share-of-voice over time.

MCP on every plan

Give your AI tools ears on LinkedIn

Add your keywords, connect the MCP server, and ask your assistant what LinkedIn is saying about you — all inside the 7-day free trial.

No credit card required · API access on all plans · Cancel anytime