Automating LinkedIn Growth with Claude Projects and Skills

Duncan Rogoff | AI Automation2026-05-22go watch the original →

the gist

A four-level framework for scaling LinkedIn content, moving from basic chat prompting to autonomous browser-based agents that research, write, and publish posts.

Scaling Content via Claude Levels

The author outlines a progression for using Claude to automate a personal brand on LinkedIn, moving from manual prompting to fully autonomous agents. The system relies on four levels of increasing complexity: basic chat, project-based knowledge management, browser-based skill execution, and a centralized knowledge vault.

Implementation Techniques

Level 1 (Basic Prompting): Use the built-in research toggle in Claude Chat to pull real-time web data into posts. The author notes that including hard statistics in the opening hook significantly improves engagement.
Level 2 (Claude Projects): Create a dedicated project to store system instructions and context files. The system prompt should explicitly define a multi-step workflow: research the topic, draft a hook, integrate personal proof points from uploaded files, and format the output.
Level 3 (Claude Code Skills): Install the Microsoft Playwright CLI to enable browser automation. This allows Claude to navigate to LinkedIn, log in, and interact with the UI. The author uses this to automate the posting process entirely.
Level 3.5 (Scheduling): Utilize the /schedule command or the routine manager within Claude Code to trigger the LinkedIn skill at specific intervals, such as 8:00 a.m. daily, cycling through a predefined list of niche-relevant keywords.
Level 4 (Knowledge Vault): Connect the agent to a local knowledge base (e.g., an Obsidian vault) containing personal stories, testimonials, and past performance data. This allows the agent to identify trending topics on Reddit relevant to the user's Ideal Customer Profile (ICP) and draft posts that align with past successful content.

Context

The author transitioned from 1,200 to 14,000 LinkedIn followers in 90 days by shifting from manual content creation to an AI-driven system. The primary bottleneck addressed is the disconnect between generic AI output and a specific personal brand voice. By moving beyond simple chat interfaces into Claude Code and browser automation, the author eliminated the need for manual copy-pasting and scheduling.