Claude · Module 06 of 8

Hidden Features

Capabilities most people miss — Artifacts, Projects, MCP, and Claude Code.

4 lessons16 exercises

Topics & techniques

Artifacts

Turn answers into editable, runnable side-panel documents.

4 cards →

Build It in an Artifact, Then Iterate in Place

Using Claude Artifacts — the side-panel where code, documents, and small apps render live — to build something and then refine it in place, instead of re-asking for a fresh copy each time.

L2 · Context engineering

Iterate Inside an Artifact Instead of Re-Pasting in Chat

A workflow for building something concrete — a document, a component, a script — where you have Claude create it as an Artifact and then refine it in place with targeted requests, rather than pasting the whole thing back into chat each round to be regenerated.

L2 · Context engineering

Turn an Artifact Into a Link You Can Share

Using Claude Artifacts to produce something you can hand to another person as a live link — a small interactive tool, a calculator, a dashboard, a formatted document — instead of copying code out and hosting it yourself. The artifact is both the workspace and the deliverable.

L2 · Context engineering

Turn a Vague Idea Into a Visual You Can Poke At

Using an artifact to make an abstract idea concrete and interactive — a chart of your data, a clickable mock of a UI, an interactive diagram of a process, a what-if calculator — so you can react to something you can see instead of arguing with it in prose.

L2 · Context engineering

Projects

Persistent context and knowledge across chats.

4 cards →

One Project per Context, Not One for Everything

Organising Claude Projects around distinct bodies of work — one per client, product, course, or ongoing topic — so each project's knowledge and custom instructions describe exactly that context, rather than piling everything into one general-purpose project.

L2 · Context engineering

Prune Project Knowledge So It Stays Sharp

Treating a Claude Project's knowledge as something you maintain, not just accumulate: periodically removing outdated files, reconciling documents that now contradict each other, and cutting material that's no longer relevant — so the project keeps giving accurate context instead of stale context.

L2 · Context engineering

Put Standing Context in a Claude Project

Using a Claude Project to hold the context you'd otherwise paste every time — reference docs, instructions, style, and background — so every chat in the project starts already informed.

L2 · Context engineering

Put Your Source of Truth in Project Knowledge

A Projects habit: load the stable reference material a body of work depends on — style guide, product facts, key docs, decisions — into the project's knowledge once, so every chat in that project starts already knowing it, instead of you re-pasting the same context each time.

L3 · Workflows

MCP

Connect Claude to tools and data with the Model Context Protocol.

4 cards →

Build a Custom MCP Server for Your Own Stack

Writing your own MCP server to expose an internal system — a private API, an in-house database, a bespoke workflow — to Claude as a small set of well-described tools, when no off-the-shelf server covers what you need. It's how you connect Claude to the parts of your stack that are unique to you.

L4 · Multi-AI systems

Connect Claude to Your Tools With MCP

Using the Model Context Protocol (MCP) to connect Claude to real tools and data — files, databases, APIs, internal systems — through a standard interface, so it can act on your actual context instead of only what you paste.

L4 · Multi-AI systems

Scope MCP Access to Least Privilege

A security discipline for MCP: grant each connected server the narrowest access that its job actually requires — read-only when you won't be writing, scoped to specific directories or datasets, no destructive capabilities it doesn't need. The connection's permissions should match the task, not the maximum the server can offer.

L4 · Multi-AI systems

Start With One MCP Server That Earns Its Place

An adoption strategy for the Model Context Protocol: instead of connecting every available server at once, wire up a single MCP server that removes a real, repeated pain in your workflow, prove it earns its keep, and only then add more. Start where the payoff is obvious.

L4 · Multi-AI systems

Extended thinking

Let Claude reason at length before answering — and know when it helps.

4 cards →

Give Thinking Room for Math, Logic, and Constraints

Turning extended thinking on for the class of problems that reward working through steps: multi-step arithmetic, logic and constraint puzzles, unit conversions, anything where a single wrong intermediate step silently poisons the final answer. These are the tasks where 'show the work' actually changes the result.

L3 · Workflows

Read the Reasoning, Not Just the Answer

A habit of actually reading Claude's extended-thinking trace — not just the final answer — to spot where its reasoning took a wrong turn: a misread requirement, a bad assumption, a step skipped. The visible reasoning is a debugging surface for the answer.

L3 · Workflows

Tell Thinking What a Good Answer Must Satisfy

Giving Claude an explicit definition of what a correct answer must satisfy — the constraints, the acceptance criteria, the must-haves — before it starts thinking, so its reasoning has a target to steer toward and check itself against, rather than optimising for 'a reasonable-sounding answer'.

L3 · Workflows

Save Extended Thinking for Genuinely Hard Problems

A judgement call about when to turn on extended thinking — the mode where Claude reasons at length before answering. Reach for it on problems that genuinely need multi-step reasoning and where a wrong answer is costly; skip it on simple or easily-checked tasks where it just adds latency.

L3 · Workflows

Claude hidden features

The features that change what Claude can do

Beyond the chat box, a handful of Claude features quietly expand what is possible. Artifacts turn an answer into a live document, component, or mini-app you refine in place. Projects hold shared context and files across many conversations. Extended thinking lets Claude reason visibly before answering hard problems. And the Model Context Protocol connects Claude to your own tools and data through a standard interface — the doorway to grounded, real-world workflows.

Match the feature to the job

Each earns its place in a different situation. Build in an Artifact when you want to iterate on something concrete rather than reading it back in chat. Use Projects for engagements that span sessions and would otherwise fragment. Turn on Extended thinking when a problem genuinely needs careful reasoning. And reach for MCP when you want Claude to act on your real data — files, databases, internal systems — instead of pasted snippets. Together they turn Claude from a chat window into a workbench.

The hidden-feature topics to explore

This section covers Artifacts, Projects, Extended thinking, and the Model Context Protocol. Pick the card below for the feature that matches how you want to work.

Frequently asked

What are Claude Artifacts?

Artifacts turn an answer into a live, iterable object — a document, component, or small app — that you refine in place rather than rebuilding in chat. They are ideal when you want to iterate on something concrete instead of reading it back.

What is Claude's extended thinking?

Extended thinking lets Claude reason step by step, visibly, before giving its final answer. It is worth enabling for genuinely hard problems where careful reasoning improves the result, and less necessary for quick, simple tasks.

What is MCP (Model Context Protocol)?

The Model Context Protocol is an open standard for connecting Claude to external tools and data — files, databases, APIs, internal systems — through one consistent interface. It is what lets Claude act safely on your real data rather than pasted text.

What are Claude Projects?

Projects hold shared context and files across many conversations, keeping a long engagement coherent instead of scattered across separate chats. They are ideal for ongoing work you return to over multiple sessions.

Test yourself

Do you know Claude's hidden features?

Six questions on the capabilities most people never touch — Artifacts, Projects, MCP, and extended thinking. No sign-up, instant score, and you can share how you did.

Q1 You've built something in an Artifact and want a variation. What's the right move?
Q2 How should you organise Claude Projects?
Q3 What do most people get wrong about Project knowledge?
Q4 What is MCP (the Model Context Protocol) for?
Q5 You're connecting an MCP server that only ever needs to read a dataset. What access do you grant?
Q6 When is extended thinking genuinely worth turning on?

0 / 6 answered