Claude Skills for LinkedIn: Automate Outreach and Prospecting Workflows

LinkedIn is still the highest-signal channel for B2B prospecting in 2026. A connection request with a personalized note gets accepted at 30–40% when personalized versus 10–15% for blank requests. The bottleneck is not the channel — it is the manual work required to send 20 well-researched, personalized requests per day.

Claude skills for LinkedIn solve this. They are modular automation workflows — built with Claude Code and a Playwright-controlled browser — that handle the research, personalization, enrichment, and sequencing that turns LinkedIn into a scalable outreach channel.

This guide covers how those skills work, what each one does, and how to configure them for your ICP. SDRs and BDRs are already using them to build pipeline in a fraction of the manual time.

Overview

LinkedIn automation has a reputation problem. Most tools that automate connection requests or message sequences get accounts restricted because they violate rate limits, use fake browser fingerprints, or ignore LinkedIn's anti-automation detection.

Claude skills take a different approach. They use Playwright to control a real browser session — your logged-in LinkedIn account, running in a headed Chrome window — and execute every action a human would take, at human speed, within the limits a human could plausibly hit in a workday.

The result is automation that behaves like an efficient SDR, not a bot. It researches the prospect, drafts a message, sends the request, and logs the interaction — staying inside the daily caps that keep your account safe.

This guide is for B2B sales teams and GTM engineers. It covers four core skills, the data layer that powers them, and configuration options for different outreach strategies.

What Are Claude Skills?

Claude skills are modular, single-purpose automation instructions that Claude Code loads and executes on demand. Think of each skill as a trained procedure: a defined input, a sequence of steps Claude follows, and a structured output.

For LinkedIn, a skill might be: "Given a list of 50 LinkedIn profile URLs, send each a connection request with a personalized note referencing their most recent post — but stop if today's request budget is exhausted." Claude handles the browser navigation, message generation, budget tracking, and output logging without any additional instructions from the user.

Skills differ from one-off prompts in three ways. They are repeatable — the same skill runs identically every session. They are stateful — each skill checks logs to know how many actions were already taken today. And they are composable — the output of an enrichment skill feeds directly into an outreach skill without manual data transfer.

The Outreach Skill: Connection Campaigns on Autopilot

The outreach skill runs an 8-stage workflow that takes an ICP definition as input and produces a batch of sent connection requests as output.

1. ICP discovery: Claude receives criteria — industry, company size, seniority level, geography, tech stack — and translates them into a LinkedIn search filter set.
2. Company list extraction: Playwright navigates LinkedIn Company Search and extracts matching companies up to the batch size limit.
3. Decision-maker identification: For each company, the skill navigates to People results filtered by seniority (e.g., CEO, CTO, Head of Sales) and extracts profile URLs.
4. False-positive filtering: Claude cross-checks each profile against the ICP — removing obvious mismatches like consultants using the target title, or profiles with less than 6 months in the role.
5. Enrichment pull: Each profile is passed to the enrichment skill to fetch recent LinkedIn activity, current company signals, and tech stack data.
6. Message generation: Claude writes a unique connection note per profile, referencing the most relevant signal (see the personalization section below).
7. Request sending: Playwright navigates to each profile, clicks Connect, inserts the note, and sends — with human-speed delays between actions.
8. Log update: Every sent request is written to a structured log with timestamp, profile URL, company, note sent, and today's running budget count.

ICP:
  industry: SaaS, B2B software
  headcount: 50–500
  titles: [VP Sales, Head of Revenue, Chief Revenue Officer]
  geography: United States
  signals: hiring SDRs in last 30 days
batch_size: 20
message_style: reference_recent_post

The "hiring SDRs" signal is especially powerful. A company actively recruiting sales development reps is investing in outbound — which means budget exists and the decision-maker is already thinking about pipeline efficiency. This is exactly the trigger that makes a Claude Code outbound sales workflow relevant at the exact right moment.

AI Message Personalization at Scale

LinkedIn connection notes have a 300-character limit. Every word matters. Generic notes ("I'd love to connect and learn more about your work") hit 10–15% acceptance. Notes that reference something specific about the person consistently hit 30–40%.

Claude generates personalized notes by ranking available signals and building around the most relevant one. The hierarchy mirrors what works in cold email personalization: personal activity beats company event, company event beats industry reference.

For accepted connections that do not reply within 5 days, the skill generates a follow-up direct message — again personalized, shorter, with a single soft ask. The follow-up maintains the same signal reference from the connection note so it reads as a natural continuation, not a fresh cold pitch.

Teams already running Claude Code LinkedIn outreach with Salesforge MCP report 2–3x higher reply rates on personalized notes versus template-based sequences using traditional automation tools.

Contact Enrichment Workflows

LinkedIn profiles contain enough to personalize a message. They rarely contain enough to build a complete sales record — verified email, direct phone, firmographic data, and intent signals.

The profile enrichment skill bridges that gap. Given a batch of LinkedIn URLs, it extracts publicly visible data from each profile — role, company, location, years of experience, mutual connections — and passes each record to SyncGTM's enrichment layer for deeper data:

• Verified business email — waterfall enrichment across 50+ providers until a deliverable address is found
• Direct phone number — mobile or direct line when available
• Firmographic signals — headcount, revenue estimate, recent funding, industry classification
• Tech stack — tools the company actively uses, detected from job postings and website signals
• Job change alerts — flag contacts who recently changed roles (high buying intent within 90 days of job change)

Output options: structured CSV for upload to your CRM, JSON for programmatic processing, or direct CRM write via HubSpot or Salesforce MCP connection.

A typical enrichment run on a 50-profile batch completes in 8–12 minutes and returns records with 85–95% email coverage — comparable to running the same list through dedicated enrichment tools at a fraction of the per-record cost.

Feed Engagement and Warm-Up

Two tactics consistently lift LinkedIn outreach performance: engaging before connecting, and targeting people who already signaled interest in your topic by commenting on relevant posts.

The Feed Engagement Skill

Before sending connection requests to a prospect, the feed engagement skill scrolls LinkedIn's feed, identifies recent posts from ICP-matching profiles, and posts a value-adding comment on their content. The next day, when the connection request arrives, the prospect already recognizes the name — and acceptance rates climb significantly.

Claude generates comments by reading the post content and producing a response that adds a specific point, asks a follow-up question, or cites a complementary data point. Not "Great post!" — that hurts credibility. A comment that shows you actually read and thought about the content.

High-Intent Lead Extraction

The high-intent skill takes a different angle: given a LinkedIn post URL on a topic relevant to your ICP, it scrapes everyone who commented or reacted, filters against ICP criteria, enriches the matches, and scores them against your ideal customer profile.

The resulting list is the highest-intent lead source available on LinkedIn. These are people who publicly demonstrated interest in a problem your product solves — before you said a word to them. Connection notes that open with "saw your comment on [post]" regularly achieve acceptance rates 15–20 percentage points above cold outreach to the same titles.

Rate Limits and Account Safety

LinkedIn's automation detection looks for two things: volume anomalies and non-human behavior patterns. Claude skills handle both.

Volume is managed through per-skill daily caps enforced by reading the action log before executing. If the log shows 20 connection requests already sent today, the outreach skill exits cleanly rather than continuing. The caps are conservative by design — they stay well below what LinkedIn flags as suspicious.

Behavior patterns are handled by Playwright running in a real browser with:

• Human-speed delays (2–5 seconds between actions, not milliseconds)
• Randomized scroll depth and dwell time on each profile
• Your real browser fingerprint (not a headless browser signature)
• Real mouse movement trajectories, not instant teleportation to elements

For teams running multiple LinkedIn accounts, each account gets its own browser profile and independent skill execution — daily caps are tracked per profile, not per team.

SyncGTM as the Data Layer

Claude skills for LinkedIn are useful on their own. They become far more powerful when every profile is backed by enrichment data from outside LinkedIn.

SyncGTM's MCP gives Claude Code a single connection to 50+ enrichment providers. When the outreach skill identifies a prospect, it immediately pulls:

• Verified email — so the LinkedIn connection is paired with a parallel cold email sequence
• Buying signals — job postings that indicate scaling, recent funding, tech stack changes
• Job change alerts — contacts who just started a new role (90-day window = highest buying intent)
• CRM status — check if the prospect is already in HubSpot or Salesforce before sending, avoid duplicate outreach

Without this layer, a LinkedIn skill is operating with only what is visible on the profile. With SyncGTM, every profile becomes a complete prospect record — enriched, scored, and CRM-ready before the connection request is sent.

Setup takes under 10 minutes: install the SyncGTM MCP, authenticate with your API key, and all skills immediately gain enrichment access. No individual API integrations for each data provider. No credential management across multiple tools.

For teams already running Claude Code sales automation workflows, adding the LinkedIn skills creates a unified prospecting pipeline: enrichment feeds outreach, outreach logs feed CRM, CRM triggers follow-up sequences — all orchestrated by Claude Code without manual handoffs between tools.

Conclusion

Claude skills for LinkedIn are not about automating mass outreach. They are about making every LinkedIn interaction higher quality — better researched, better personalized, and better timed — while eliminating the hours of manual work that keeps most SDRs from running LinkedIn as a consistent pipeline channel.

The key advantages over traditional LinkedIn automation tools: real browser execution that avoids detection, AI-generated personalization that beats templates, and enrichment integration that gives each prospect context beyond what LinkedIn shows. Together, those three things shift the output from volume to quality.

Start with the outreach skill. Define your ICP, run a 20-person batch, and measure the acceptance rate against your last manual campaign. The personalization difference shows up immediately. Add the enrichment and feed engagement layers after you have the core workflow running. The full pipeline — outreach to CRM-ready lead — takes under two hours to configure for a first deployment.

For teams using LinkedIn for B2B sales at scale, Claude skills turn the highest signal-to-noise channel in B2B prospecting into one that actually scales.