Claude Hub
Every Claude workflow I've tested and published: Code workflows, Cowork setups, Skills tutorials, CLAUDE.md files. Real prompts, real failures, nothing theoretical.
Last updated: June 2026
TL;DR: Claude Hub is Product with Attitude’s independent Claude guide library for builders who want to use Claude for real work. It covers Claude Chat, Cowork, Design, Code, Skills, prompts, CLAUDE.md, connectors, memory, long-context workflows, and the major 2026 Claude releases. Start with Chat for questions, Cowork for autonomous workflows, Design for visual drafts, Code for terminal-level development, and Skills or CLAUDE.md when you want Claude to stop forgetting how you work.
I started writing about Claude in September 2025.
Back then, Cowork didn’t exist, Skills weren’t a thing, and Code was a niche CLI tool that maybe a few hundred developers used daily.
Many posts and tens of thousands of words later, this page exists to help you decide where to start.
This is everything I’ve published about Claude on Product with Attitude, organized so you can find what you need and skip what you don’t.
Every guide is tested. Every prompt is real. Every failure is documented. Nothing theoretical.
This page is also intentionally structured as a 2026 Claude guide.
Anthropic’s documentation tells you what each Claude feature does. This library tells you which surface to use, what breaks, which prompts survive contact with reality, and where the workflow starts paying rent.
If you only remember one thing: Claude is no longer one chatbot. It’s an ecosystem. And choosing the right Claude surface for the job is a skill work developing.
Hey, I’m Karo Zieminski 🤗
AI Product Manager and builder.
I write Product with Attitude, an AI newsletter for thousands of subscribers developing critical AI literacy the only way it sticks: through practice.
We don’t just use AI. We build workflows, automations, and products with it, while studying how AI itself is built, positioned, and woven into our work.
If you’re new here, welcome!
What’s Inside
What is Claude Hub?
Claude Hub is Product with Attitude’s independent Claude guide library for builders, PMs, solo founders, creators, and AI practitioners who want to use Claude for actual work. It’s not a rewrite of Anthropic’s documentation. It’s the field manual after the docs: tested workflows, working prompts, broken assumptions, practical tradeoffs, and the “don’t waste three hours on this” notes you only get by building with the thing.
The library covers Claude Chat, Claude Cowork, Claude Design, Claude Code, Claude Skills, CLAUDE.md files, connectors, prompts, long-context workflows, and Claude’s 2026 release cycle. It exists because Claude has become too big to navigate from memory. Which is annoying. Also powerful.
Use this page when you need a Claude guide for 2026 and want to answer one practical question first: which Claude surface should I use for this job?
Claude Guide 2026: The Fast Decision Map
Claude is Anthropic’s AI assistant, but in 2026 the product is not one app but a stack of working surfaces. Chat, Cowork, Design, Code, Skills, Memory, Projects, connectors, and CLAUDE.md all solve different problems. Treating them as interchangeable is how people end up with messy AI habits.
Claude Cowork vs Chat vs Design vs Code
Claude is Anthropic’s AI assistant. Chat, Cowork, Design and Code are four different ways to use it. Choosing the right mode depends on what you’re trying to accomplish.
Which One Do You Need?
Chat is a conversation. Cowork is a workflow engine. Design is a design partner. Code is a development partner.
That’s the simplest way we can frame the three modes for anyone deciding where to start.
The short version: use Chat when the work lives in the conversation, Cowork when the work lives across files and tools, Design when the output needs visual structure, and Code when the work lives in a repo.
Use Skills and CLAUDE.md when you want the behavior to repeat without re-explaining yourself like a cursed onboarding manager.
Claude Releases 2026
Claude changed materially in 2026. The old “best Claude prompts” format is now too small for what the product became.
My read: the biggest 2026 Claude shift is product unbundling. One model launch did not define the year; the new working surfaces did.
In January, Anthropic moved Cowork from idea to research preview, first for Max users on Claude Desktop, describing it as Claude Code-style agentic capability for knowledge work beyond coding.
Cowork moved Claude beyond chat and toward workflow execution, with access to files and MCP integrations.
My PM read: Chat requires users to package context manually. Cowork reduces that burden by giving Claude access to the materials and integrations that already define the workflow. That changes adoption. The user no longer has to translate work into prompts as much.
My critical AI literacy read: This is also where risk increases. Once Claude can work around files and MCP integrations, quality is only one question. The harder questions are access, permissions, logging, approvals, and failure paths.
In February, Anthropic added scheduled tasks in Cowork, grouped Skills, plugins, and connectors into a new Customize section in Claude Desktop, launched a Cowork plugin marketplace and admin controls, introduced Claude for PowerPoint, improved Claude for Excel, and launched Sonnet 4.6 with upgrades across coding, computer use, long-context reasoning, agent planning, knowledge work, and design. Sonnet 4.6 also introduced a 1M-token context window in beta.
My PM read: This release cluster is about repeatable workflow infrastructure.
Scheduled tasks reduce manual prompting.
Customize makes capabilities easier to discover and manage. Plugins, connectors, marketplaces, and admin controls make Claude more viable for teams.
PowerPoint and Excel move it into the boring-but-expensive work where companies bleed time.
The product strategy is clear: Claude is being shaped into a managed work layer.
My critical AI literacy read: Every new connector, plugin, admin setting, and scheduled task expands what Claude can touch. That creates real value, but also more permission design, logging needs, failure modes, and accountability questions. The more Claude becomes part of recurring workflows, the less acceptable “the model got confused” becomes as an excuse.
In March, Claude became more persistent and more visual. Free users got memory from chat history, charts and visualizations appeared inline, Excel and PowerPoint context sharing improved, Cowork threads persisted from mobile, and computer use entered research preview in Cowork and Claude Code. The pattern was clear: Claude is being built for persistent, tool-using, cross-surface work.
My PM read: This is a product maturity move. Anthropic is reducing restart friction. Memory helps Claude keep context. Inline charts make outputs easier to inspect. Office context sharing brings it closer to existing documents. Persistent Cowork threads make mobile part of the workflow instead of a side door. Computer use previews point toward action. The product bet is continuity: fewer cold starts, fewer manual handoffs, more work that survives across sessions.
My critical AI literacy read: A stateless chatbot is easier to reason about because every session mostly starts fresh. A persistent, tool-using, cross-surface system creates better workflows, but also bigger questions: what is remembered, what is inferred, what is allowed to persist, what crosses between surfaces, and who notices when old context becomes wrong. Persistent AI needs memory hygiene, not blind trust.
In April, Claude Cowork became generally available on macOS and Windows through Claude Desktop. Anthropic also launched Claude Design by Anthropic Labs and released Opus 4.7, with stronger software engineering, long-running coding, and higher-resolution vision. The Chat, Cowork, Design, and Code split now has teeth. Each surface is built for a different job.
My PM read: This is product segmentation by job-to-be-done. Anthropic is not only adding features. It is giving users different surfaces for different work modes. That reduces cognitive load when done well: users know where to go for conversation, workflow operation, design work, or coding.
My critical AI literacy read: Different surfaces create different accountability models. Chat can be wrong in text. Cowork can affect workflow. Code can change software. Design can shape visual communication and brand assets. The same model family behaves differently depending on the surface, permissions, context, and output type. Treating them as one generic “Claude” experience hides the real risk profile.
In May, Anthropic launched Opus 4.8, saying it improved over Opus 4.7 in coding, agentic skills, reasoning, and practical knowledge work tasks. It also expanded enterprise controls for connectors and compliance integrations. The direction is boring in the best way: better agents, better governance, better work.
My PM read: Capability + control = enterprise AI game.
My critical AI literacy read: Better agents create more value only if teams can define what those agents can access, what they can change, and how their actions fit compliance requirements.
Claude Opus 4.8
Claude Opus 4.8: What Changed, and How I’ll Test It
In this article, I break down what changed in Claude Opus 4.8, from the honesty upgrade and effort controls to Dynamic Workflows, and argue its real significance isn’t raw power but that it’s a rehearsal model teaching us the inspectable, agent-orchestrating habits we’ll need before Claude Mythos arrives.
Claude Opus 4.7
Anthropic Launches Claude For Small Business
In this article, I break down Anthropic’s May 13, 2026 launch of Claude for Small Business: Claude Cowork, 15 agentic workflows, 15 reusable skills, and 8 connectors, including QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, Microsoft 365, and Slack.
Every other launch piece I read listed the workflows. The part they missed was implementation.
Which workflow should a small business try first? Which actions should stay read-only? Which ones need approval? And which ones should absolutely not run on autopilot because someone saw a nice product demo and briefly lost respect for accounting?
That’s what this guide covers: a pain-based decision tree, a trust-boundary table, and a four-week rollout plan that starts with read-only access before Claude touches anything expensive, legal, public, or customer-facing.
I Mapped the Opus 4.7 Release to Your Role, Goals, and Real Workflows
In this article, I break down the release of Claude Opus 4.7, launched on April 16, 2026, including pricing changes, API breaking updates, benchmark results, and what changes for builders, writers, PMs, and vibe coders.
Why Claude Design Only Works Because of Opus 4.7
In this article, I explain why Claude Design works in the first place, and why that comes down to Claude Opus 4.7, whose jump in vision accuracy and self-checking behavior makes reliable UI interpretation possible.
Claude Cowork
Cowork handles autonomous knowledge work: it plans tasks, spawns parallel sub-agents, reads and writes real files in a sandboxed environment.
Claude Cowork Guide for Power Users: 50+ Tested Tips on Plugins, Skills, Sub-Agents, and Memory
This is the big one. 50+ tested tips covering plugins, skills, sub-agents, memory, and the settings most people never touch. I spent nine days breaking Cowork before writing a single word.
What works, what breaks, and how to make Claude Cowork useful in 2026.
I Built a Claude Cowork Loop That Improves Itself. Here’s the Exact Setup.
In this article, I show you how to turn Claude Cowork into a no-code, self-improving automation loop that rewrites its own instructions after every run, so quality compounds over time without you touching a terminal.
What makes it unique is that I take Karpathy’s code-heavy autoresearch pattern from March 2026 and rebuild it entirely inside Cowork’s visual interface using just three pieces, a context.md file, folder instructions, and a Self-Improvement Directive, so even people scared of Claude Code can run an agent that gets smarter on its own.
How I Turned Notion Into Claude’s Persistent Memory, and 10+ Connector Tips Most Guides Skip
In this guide, I walk through how I set up the Cowork-Notion Connector and share 10+ tips most guides skip. Connectors are what turn Cowork from a smart assistant into an integrated workflow engine.
Anthropic Shipped Cowork in 10 Days Using Its Own AI. Here’s Why That Changes Everything.
I introduced Cowork for the first time, broke down the product decisions behind it, and explained what it means for knowledge workers. Anthropic built Cowork using its own AI in 10 days. That kind of acceleration should make product leaders nervous.
Claude Design
Launched in April 2026, it gives you a way to create clean, polished visual assets like website prototypes, slide decks, and marketing one-pagers just by describing what you want.
Claude Design Review: 48-Hour Builder’s Test + Hero Prompts
In this article, I spend 48 hours hands-on testing Claude Design running on Opus 4.7, and I show you the exact prompts I used to build animated hero sections, where the tool burns through credits fastest, and how the real product story is the Design plus Code plus research plus memory stack Anthropic is assembling, not the design tool on its own.
Instead of riding the “Figma is dead” hot take that tanked Figma’s stock 7.28 percent, I separate design production from design thinking and argue Claude Design is a first-draft killer rather than a Figma killer, since it cannot do real-time collaboration and the AI’s render delays keep breaking your creative flow.
Claude Skills
Claude Skills are reusable markdown instruction files that turn Claude from a general assistant into a specialized agent. They are the fastest way to get consistent, repeatable results from any Claude workflow.
What Are Claude Skills and How to Build Them
A step-by-step creation guide, a curated list of community skill repositories, and three ready-to-copy skills I built in under 90 minutes.
Claude Skills Library
I maintain two tested Claude Skills libraries: one with skills I’ve built myself, and one with skills shared by the Product with Attitude community.
You can find the community-built Claude Skills in Attitude Vault. Built one yourself? Share it there too.
Claude Skill: Alt Text for the AIO Era
This one is about SEO and AIO, but the Claude Skill inside it is the practical payoff. Google’s AI Overviews now parse alt text as a semantic unit alongside the image. New rules: 125-200 characters, entity-rich, describe purpose not appearance. The post includes a ready-to-copy Skill that generates optimized alt text and filenames automatically.
Claude Skill: LinkedIn Teaser Generator Skill
Turns a Substack article into a brand-voice LinkedIn teaser post with a strong hook, a first-comment link, and ready-to-use posting metadata.
Claude Skill: The Context Window Hygiene Skill
This skill works in Claude Code and Claude Cowork, so you can use it both when coding and when managing longer agentic work sessions. It helps you keep an AI agent’s working space clean, focused, and useful.
Claude Prompts
Stunning Hero Prompts for Claude Design: Recreate the Exact Design
In this article, I share the exact prompts I used to create stunning hero section designs in Claude Design.
n8n AI Agent Builder for Claude Code
How to build production-grade n8n AI agent workflows without manual JSON editing. Includes a prompt that generates fully importable n8n workflow JSON from plain-language descriptions.
Claude Code
How Boris Cherny, Builder of Claude Code, Uses It, And Why That Should Change How You Think About AI
Boris built Claude Code itself. His workflow shows how to use Claude Code as a real development tool, not a toy. Nine practices that separate productive AI-assisted development from prompt-and-pray.
CLAUDE.md
CLAUDE.md is a project instruction file for Claude. It defines conventions, context, guardrails, commands, and expectations so Claude does not waste every session relearning how your project works.
CLAUDE.md and Rules Files
Rules files encode conventions, guardrails, and product context so your AI stops improvising. The Agent Rules Generator prompt outputs platform-specific files for Replit, Cursor, Gemini, and Claude in one session, with build-now vs. build-later splits and plain-language explanations for every rule.
Why Claude Tokens Shape Your Workflow
Claude Just Unlocked 1 Million Tokens For Everyone. Here Is What That Means.
In this article, I break down Anthropic’s March 13, 2026 move to make the 1 million token context window generally available on Opus 4.6 and Sonnet 4.6, why removing the long-context pricing surcharge is a deliberate competitive jab at OpenAI’s pricing cliff, and what it changes for the way you work day to day.
Instead of doing the usual codebase-and-Claude-Code angle everyone else ran with, I translate it into plain whiteboard language, pull in the MRCR v2 recall numbers where Opus hits 78.3% versus Gemini’s 26.3%, and then map the practical payoff role by role for writers, researchers, agent builders, and PMs, while staying honest about context rot, cost explosions, and why a bigger window can punish lazy planning.
Claude in Comparisons
Perplexity Computer vs Claude Cowork vs OpenClaw
Perplexity Computer orchestrates 19+ frontier models (including Claude Opus 4.6 for reasoning) from a single prompt. The review includes head-to-head comparisons with Claude Cowork and OpenClaw.
Claude in Builder Stories
The Time Claude Said “Don’t” and the Builder Ignored It
A real case study in trusting your product judgment over AI advice. I documented every moment Claude suggested removing features, and Karen Spinner overruled it. Nobody else publishes these stories because they make the AI look fallible. That’s why I did.
Join thousands of readers learning AI the only way it sticks: through real experiments and real projects.
Cross-Tool Resources That Feature Claude
The Ultimate Vibecoding Guide
A 16-step roadmap for building production-ready apps with AI, co-authored with Karen Spinner. Claude is one of the primary tools alongside Replit, Cursor, and ChatGPT. Covers mindset, validation, debugging, testing, and cost control.
10 Tools for Running a Bestselling Substack in 2026
Claude Code and Cowork are the backbone of my daily workflow. This post shows how they fit alongside Perplexity, Recraft, and the rest.
Self-Improving Prompt System
A Prompt Builder paired with a Prompt Evaluator that scores against a 35-criteria rubric. Works across ChatGPT, Claude, Gemini, and any LLM that accepts structured instructions.
Share this with 3 friends or colleagues and you’ll get a free month of premium membership.
What Makes This Library Different
Who is this for?
Anyone who uses Claude for real work: writers, PMs, builders, creators, knowledge workers. Whether you picked up Claude yesterday or you’ve been using it since Sonnet 3.5, you'll find a workflow you haven't tried.
If people keep telling you to “just use AI,” this guide is for you. Especially if you are building solo, managing AI products, running a newsletter, serving clients, or stitching tools together with duct tape and ambition. Claude works better when you give it structure: the right surface, context, files, rules, and constraints.
How is Product with Attitude’s Claude Hub different from Anthropic documentation?
Anthropic explains what Claude can do. I show what works: the prompts, the failures, the workarounds they left out. I broke three workflows following the official docs before figuring out what was missing. That’s why these guides exist.
Official docs are the map of the product. This library is the map of the work. Different job. Different standard.
How often is Product with Attitude’s Claude Hub updated?
Every time I publish a new Claude-related guide, it gets added here. The “Last updated” date at the top reflects the most recent addition. As of June 2026, this library covers Claude Chat, Cowork, Design, Code, Skills, prompts, CLAUDE.md, connectors, long-context workflows, comparisons, and builder case studies.
This library started with one Cowork post in January 2026 and grew into one of the most comprehensive independent Claude resources on Substack. Every guide is written from the perspective of someone who builds with Claude daily, not someone who read the changelog.
FAQ: Claude Guide 2026
What is the best Claude guide for 2026?
The best Claude guide for 2026 should help you choose the right Claude surface, not just copy better prompts. Start with the decision map: Chat for questions, Cowork for autonomous workflows, Design for visual drafts, Code for development work, and Skills or CLAUDE.md for repeatable behavior.
What is the difference between Claude Chat, Cowork, Design, and Code?
Claude Chat is for conversation and quick artifacts. Claude Cowork is for autonomous knowledge work across files, tools, connectors, and tasks. Claude Design is for turning prompts into editable visual outputs. Claude Code is for terminal-level development workflows with project context.
Is Claude Code only for developers?
Claude Code is built for development work, but non-developers can still use it when they understand the project, constraints, and desired output. The catch: if you cannot explain the job, the repo, and the acceptance criteria, Claude Code will happily accelerate your confusion. Very fast. Very confidently.
What is CLAUDE.md and why does it matter?
CLAUDE.md is a project instruction file used with Claude Code to define conventions, context, guardrails, commands, and expectations. It matters because agentic coding gets much better when Claude does not need to rediscover your project rules every session.
WHY SUBSCRIBE ・YOUR BENEFITS・ TOOLS I BUILT・CLAUDE HUB・PERPLEXITY HUB ・VIBE CODING HUB