Connect Cursor to your LinkedIn data
Cursor gets your LinkedIn posts and analytics as tools inside the editor: pull your best hooks, summarize your voice, draft in it without leaving your code. Added from settings like any remote server, same for Windsurf and VS Code.
What worked on my LinkedIn this past month?
Compared to the previous 30 days:
- Impressions are up 42% and comments up 31%.
- Your 3 best posts all open with a question and stay under 120 words.
- Tuesday-morning posts outperform your average by 2.1x.
Want me to draft the next one using what's working?
Set up Cursor in under 2 minutes
Nothing to install and nothing to configure. You log in to MyFeedIn once and Cursor keeps the connection.
https://mcp.myfeedin.co1. Open Cursor settings
Cursor Settings, then MCP, then Add new MCP server. The label varies slightly by version.
mcp.myfeedin.coRemote server
2. Paste the MyFeedIn MCP URL
Add it as a remote server with the MyFeedIn server URL. No API key or JSON config needed.
Read your posts, analytics, comments, feeds and people
Approve3. Approve in your browser
Cursor opens myfeedin.co for OAuth consent. Approve, and the tools are available in chat.
What Cursor can do once connected
Four things you cannot do with a chat window alone, because the model has no idea what happened on your account.
Find out what actually worked
Ask what changed this month and get the real deltas: impressions, comments, best day, best hour, best post length. From your own history, not from a LinkedIn best-practices article.
Draft in your own voice
Your assistant reads how you actually write, your structure, your recurring words, your hooks, then drafts from that. It can also surface opening lines that worked and that you have not reused in months.
See what your commenting is worth
Nothing else measures the comments you leave on other people's posts. Ask which ones landed, on whose posts, and whether the time you spend engaging is paying you back.
Turn a conversation into a campaign
Name the people you want to reach, and your assistant builds the campaign: the audience, the sequence, the delays. It lands in your dashboard as a draft, and nothing goes out until you launch it.
Then ask Cursor anything
These work as written. Paste one in and see what comes back.
Compare my last 30 days on LinkedIn to the 30 days before. Show me the 3 metrics that moved the most (impressions, comments, reactions) with the percent change. Then pick my single best post in that window and break down what made it work: hook, length, posting time, and how it compares to my average.
Read my last 20 posts and summarize my voice in 5 bullets: average post length, opener style, formatting habits (line breaks, bullets, emoji use), recurring themes, and the 3 hook patterns that drive the most engagement. End with a 2-line guide I can paste into any AI tool to write in my style.
Pull 5 hooks from my posts that got more than 500 impressions in the past year and that I haven't reused in the last 90 days. Adapt each one to a post about [topic] without changing the hook's structure. Keep every hook under 12 words.
Find my 3 best-performing posts from 11–13 months ago. For each one, give me: the original hook, two reasons it worked, and a fresh angle to repurpose it today (not a copy-paste rewrite). Order them by potential, easiest repurpose first.
Draft a LinkedIn post in my voice about [topic]. 150 words. Structure: hook under 10 words, two beats with one concrete example each, close ending in a question. Plain text only, no markdown, no emojis. Pull the hook pattern from one of my top 5 performers and tell me which one you used.
Analyze my engagement patterns over the past 90 days: best days, best hours, and how post length correlates with engagement. Then give me a simple posting schedule for next week (day, time, suggested length) based only on my own data, not generic best practices.
List the people in my [feed name] feed, then check my recent comments: who in that feed have I not engaged with in the last 30 days? Give me the top 5 to reconnect with, and for each one a reason based on their headline.
Look at my top 10 posts from the past year and group them into 3 to 5 recurring themes, ranked by average engagement. For each theme, suggest 2 new post ideas I haven't covered yet. Format it as a 2-week content calendar.
Create a draft campaign called [name]. Sequence: a connection request with a short note using my first name variable, then a message 3 days later that opens with a spin variation and stays under 60 words. Write both in my voice, based on my last 20 posts. Then add these profiles: [paste 3 profile URLs], and tell me who was newly created versus already in my CRM, and who got skipped for being active in another campaign.
See all 25 tools the server exposes. Wondering whether there is an official LinkedIn MCP server? There is not, and here is how the options compare.
Built to be trusted
Your LinkedIn presence is your asset. Here is exactly what Cursor can and cannot do with it once connected.
It cannot post or message as you
No tool publishes a post, writes a comment or sends a message. The only writes are the Growth-plan campaign tools.
No scraping, no session cookie
It never touches your LinkedIn session. Only the data already in your MyFeedIn account, nothing scraped from LinkedIn.
You approve every connection
You pick whether it sees your personal account or one workspace. It can never see the other.
Revocable in one click
Settings, then Connected apps, lists every assistant. Revoking cuts it off immediately.
Frequently asked questions
Can't find the answer you're looking for? Send us a message and we'll help you out.
Also works with