AI & Product Management | Founder, Author @ The Product Compass Newsletter | Join 119,000+ PMs

Follow me for actionable tips, resources, and step-by-step playbooks for AI and AI-empowered PMs. 

The Product Compass Newsletter trusted by 125,000+ PMs: https://www.productcompass.pm/

---

I can no longer accept connections because LinkedIn limits the number of 1st-level connections to 30K. 

To connect with me, join our premium Slack community: https://www.productcompass.pm/p/product-management-resources, or engage with my content.

Paweł Huryn

1,100+ GitHub stars in 3 days — I didn't expect this.

I built an open-source PM Skills Marketplace for Claude: 100+ skills and commands that turn AI into a product management partner.

Not generic prompts. Structured skills that actually know PM frameworks.

8 plugins covering the full PM lifecycle:

→ Product Discovery: problem & solution space exploration
→ Product Strategy: vision, business model, strategy docs
→ Execution: PRDs, OKRs, roadmaps, sprint planning, user stories
→ Market Research: personas, segmentation, TAM/SAM/SOM
→ Data & Analytics: cohorts, A/B testing, retention analysis
→ Go-to-Market: GTM strategy, beachhead, ICP, growth loops
→ Marketing & Growth: positioning, North Star Metric
→ PM Toolkit: including PM resume review

Built for Claude Code and Cowork. Compatible with Gemini CLI, Cursor, Codex CLI, and Kiro.

Repo URL + instructions for Claude Code & Cowork: https://lnkd.in/dxdTj24P

Want to contribute? The repo is open. Check CONTRIBUTING[.]md.


For more about Cowork, see the Ultimate Guide for PMs: https://lnkd.in/dyY-jhvG

AI PM | Deep research. I build, test, then teach | 130K+ subscribers

Anthropic's plugins wiped $285 billion from legacy software stocks in a single day. They're now sitting in your Cowork sidebar.

Now available on Windows and Intel-based macOS.

Most people think Cowork is an improved Chat. It's not. It can do any knowledge work Claude Code can — without a terminal and with a better UX.

𝗪𝗵𝗼 𝗲𝗮𝗰𝗵 𝘁𝗼𝗼𝗹 𝗶𝘀 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗳𝗼𝗿:

→ Chat — conversations, artifacts, quick answers. No file access, no sub-agents, no plugins.

→ Cowork — autonomous knowledge work. Plans tasks, spawns parallel sub-agents, reads and writes real files, sandboxed VM. For PMs, legal, marketing, ops.

→ Code CLI — same power as Cowork, direct filesystem access, project-scoped config. For developers and engineers.

𝗛𝗼𝘄 𝘁𝗼 𝘀𝘁𝗮𝗿𝘁 𝗶𝗻 𝟱 𝘀𝘁𝗲𝗽𝘀:

1. Install Claude Desktop
2. Switch to the Cowork tab
3. Enable connectors and plugins
4. Select your working folder
5. Describe your task — done

𝗤𝘂𝗶𝗰𝗸 𝗴𝗹𝗼𝘀𝘀𝗮𝗿𝘆:

→ Skill — a reusable instruction set that teaches Claude a specific behavior 
→ Plugin — a bundle of skills + MCPs tailored to a business role
→ Connector — connects Claude to external or local apps via MCP

𝗧𝗼𝗽 𝟳 𝘀𝗸𝗶𝗹𝗹 𝗮𝗻𝗱 𝗽𝗹𝘂𝗴𝗶𝗻 𝗺𝗮𝗿𝗸𝗲𝘁𝗽𝗹𝗮𝗰𝗲𝘀:

→ github[.]com/anthropics/skills — official repo. Document skills plus creative, technical, and enterprise examples

→ github[.]com/anthropics/knowledge-work-plugins — Cowork's default plugin registry. The 11 business-role plugins

→ github[.]com/anthropics/claude-code — Developer-focused workflows.

→ claudemarketplaces[.]com — A marketplace of marketplaces

→ github[.]com/travisvn/awesome-claude-skills — Battle-tested skills for TDD, debugging, collaboration

→ github[.]com/sickn33/antigravity-awesome-skills — 868+ universal agentic skills. Role-based bundles: Startup Founder, Marketing & Growth, and more

→ skills[.]sh — Product strategy frameworks, pricing strategy, launch playbooks, interviews, PRD generator, analytics, resume optimizer

𝗪𝗵𝗲𝗿𝗲 𝘁𝗼 𝗳𝗶𝗻𝗱 𝗠𝗖𝗣 𝘀𝗲𝗿𝘃𝗲𝗿𝘀:

→ github[.]com/modelcontextprotocol/servers: Official MCP server repo — filesystem, GitHub, Google Drive, Slack, and more

→ modelcontextprotocol[.]io/examples: Official MCP directory — reference MCP implementations

→ github[.]com/punkpeye/awesome-mcp-servers: Community-curated list — hundreds of MCP servers by category

→ mcp[.]so: MCP server registry

𝗕𝗼𝗻𝘂𝘀: 𝟮 𝘁𝗵𝗶𝗻𝗴𝘀 𝘁𝗵𝗮𝘁 𝟮𝘅 𝗖𝗼𝘄𝗼𝗿𝗸'𝘀 𝗽𝗼𝘄𝗲𝗿:

→ Add Desktop Commander — full laptop access, file editing, installing MCPs.

→ Give Cowork cross-session memory — Global Instructions + Desktop Commander = memory that survives crashes and new sessions.


I broke it all down in my free guide. 

No email. No paywall: 
https://lnkd.in/dyY-jhvG

---

P.S. Want to break into AI PM? I recommend the AI PM Certification by leaders shipping in the top AI companies

A $500 discount + 1 year subscription to my newsletter with this link: https://bit.ly/aipmcohort

AI PM | Start building. Stop theorizing. I build with AI and share what works. 129K+ subscribe.

How to master Claude Cowork as a Product Manager 

Most people use Claude like a chatbot.

That's like using Photoshop only to crop photos.

Same models. Three surfaces. Very different results:

Chat → Thinking & Content
You type, Claude answers. No files, no workflow.

Cowork → Files & Workflows
Give Claude your files, set a task, step away.
Come back to real .docx/.pptx/.xlsx in your folder.

Code → Code & Systems
VS Code, terminal, desktop app, or browser.
No sandbox — git, bash, subagents, full workspace.

---

"I want to..."

Chat:
→ Draft an email or rewrite a paragraph
→ Roleplay as my VP of product 
→ Summarize a 30-page report

Cowork:
→ Research a topic — finished .docx in my folder
→ Extract tables from 50 PDFs
→ Draft replies to unreplied Slack DMs daily

Code:
→ Turn a PRD into a working prototype
→ Start on laptop, continue from phone
→ Run a long task with zero local setup (web)
→ Build systems that compound

---

What most people miss:

☑ Power Combo: Cowork /schedule for persistent daily cron that survives restarts. Code /loop for in-session monitoring. Same connectors. You can use both.

☑ Work From Anywhere: Start in VS Code, continue from your phone via /remote-control. Or kick off a long task on claude[.]ai/code with zero setup.

☑ Shared DNA: Same models (Opus, Sonnet, Haiku). Project instructions in Chat, CLAUDE[.]md in Cowork + Code. Web connectors, skills, extended thinking — all three. 

☑ Bonus: Real file access + shell commands — all three after installing Desktop Commander MCP for Claude Desktop. Switch mid-task when needed.

---

Free skill repos:
→ Knowledge work: https://lnkd.in/d6ZJbET3
→ Antigravity: https://lnkd.in/dYDzgYK6
→ Awesome Claude: https://lnkd.in/dEe8jcGa
→ Gemini: https://lnkd.in/d5dieggp
→ Product Management: https://lnkd.in/dxdTj24P

Free courses by Anthropic:
→ Claude 101 — https://lnkd.in/davAeVzv
→ Claude Code in Action — https://lnkd.in/dnK6MmnG
→ AI Fluency: Foundations — https://lnkd.in/dtvK8k2E
→ Introduction to MCP — https://lnkd.in/dU86EUTe
→ Introduction to Agent Skills — https://lnkd.in/dN2XTXSP

---

Not everyone needs Code.
But if you used Cowork, you're 70% there.

Using Code doesn't mean you stop using the others.

Pick the one that matches your task.

I wrote two full guides:

👉 Claude Cowork Guide for PMs: https://lnkd.in/dyY-jhvG
👉 Claude Code Guide: How to Transition: https://lnkd.in/dVFJ4Bix

---

By the way, if you want to break into AI PM, I recommend the AI PM Certification. A special $500 discount + 1-year subscription to my newsletter with this link: https://bit.ly/aipmcohort

JUST IN: Meet Code Review by Anthropic. It hunts for bugs at PR time. Loved it.

But security isn't just about reviewing changes. It's about stress-testing what's already there.

Here's how I approach it with Claude Code:

First, prepare. Before any security prompt works, Claude needs context:
→ Document your system architecture 
→ Document roles and permissions 
→ Document scheduled work and email flows 
→ Document secrets and variables

Then, stress-test your full codebase:
→ Someone just accessed a user's email. Figure out how. 
→ A user called an endpoint and got data they shouldn't. Trace it. 
→ User-submitted content executed JavaScript and stole admin cookies. 
→ An attacker tries to bypass access control — JWT manipulation, session hijacking, cross-tenant access. 
→ Compare what permissions.md says vs. what the API actually enforces. 
→ Audit Content Security Policy for weak directives.
→ Review against OWASP top 10.

The pattern: describe a breach scenario. Claude reverse-engineers how it could happen in your codebase. Outputs risk level, attack path, impact, and fix.

No coding required.

Run before launch, after major changes, or /loop into your pipeline.

You're welcome!

taplio

The Battle

Paweł Huryn

Ramunas Norvaisas

Followers

Posts

Engagements

Likes

Performance ScorePaweł Huryn's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Paweł Huryn

Ramunas Norvaisas

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Paweł Huryn's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScorePaweł Huryn's LinkedIn Performance