Score
9.0
/ 10
Installs
0
Repo Stars
—
Last Updated
—
Quality Ratio
90%
Description
Verified
First Published
Apr 2026
Skill Definition
Identify high-intent prospects for the user's product/service, send connection invites on LinkedIn, then check for recently accepted connections and send follow-up messages. Log everything for tracking.
Target: The user specifies their ideal customer profile: job titles, company size range, pain points their product solves, and preferred geography. Map these to specific LinkedIn search terms before starting.
Process (In one single browser automation session, so only one task/subtask, not several):
Stage 1: Find Prospects & Send Connection Invites
-
Find potential leads on LinkedIn. Go to people search at https://www.linkedin.com/search/results/people and search for leads matching the target profile. Collect their profile data and URLs. Don't make searches too complex: you can go a long way by combining a string of keywords (target job titles, industry terms, pain-point keywords, etc.) and setting the filter at the top to 2nd connection and the user's preferred geography (if the user specifies a region like "EU" that LinkedIn doesn't accept as a single country filter, use individual countries). Try different keyword combinations. Often you don't even need to open the filter sidebar. Do not get off track by trying to find details like company size; if it's not readily visible on the profile, just accept it and move on.
-
Send connection invites to 10 suitable leads you find (without a note).
-
When you have sent the connection invites, log the found leads in the current already existing CSV file in your internal files called prospect_leads.csv (create it if it doesn't exist). Columns: Lead #, Name/User, Profile Link, Job Title, Company, Specific Pain Point Criteria, Product Feature Mapping, Outreach Message Draft. The "Outreach Message Draft" column should contain a personalized follow-up message draft ready to send once the connection is accepted.
Stage 2: Check Accepted Connections & Send Follow-Up Messages
-
After completing the invites above, go straight to https://www.linkedin.com/mynetwork/invite-connect/connections/ and check for accepted connection requests ONLY from the last 24 hours (the ones accepted yesterday/today). Do not scroll further or search beyond the most recent connections. These are people who just accepted your invite.
-
For every connection accepted within the last 24 hours that hasn't been sent a message yet, send a personalized follow-up message. Check the message thread first: if there are already messages in the chat with that person, SKIP them entirely. Use the outreach message drafts from the CSV if available, or craft a new personalized message based on their profile.
RULES:
- ALWAYS check for existing message history before engaging with any prospect. If there are already messages in the chat with that person, SKIP them entirely. This is critical: sending connection invites or messages to people you already have an ongoing conversation with will interfere with those conversations. No exceptions.
- Focus on the lead themselves and their LinkedIn profile. Don't get stuck on necessarily trying to find all company info of that lead.
- Don't get stuck on trying to fill the entire sheet with lead data. It's fine if data is not always entirely complete. The most important information is name, profile link, job title, and company.
- You can do all of this in one browser automation session. No need for several sessions.
- Do not get off track. When you find an interesting lead, send the connection invite right away.
- The CSV should accumulate across sessions. Append new leads to existing data, do not overwrite previous entries.
- The outreach message drafts should be personalized to each lead's specific role and pain points, not generic templates. Reference something specific from their profile.
- Ask the user for their product/service, value proposition, target persona (titles, company size, geography), and key pain points their product addresses before starting, if not already specified.