Hidden Features
Learn artifacts, Projects, connectors and the effort control, and what each one is for.
Artifacts
Build documents, tools and prototypes in a panel beside the chat, edit them in place, and share them with a link.
4 techniques →-
How Do You Build Something in a Claude Artifact? Ask Claude for a self-contained thing, such as a document, slide deck, small tool or dashboard, and it opens in a panel beside the chat. Level 2: Context
- 1Ask for a finished thing
- 2It opens beside the chat
- 3Edit it or ask for changes
- 4Export or share it
-
How Do You Edit a Claude Artifact Without Starting Over? Change one thing at a time and say exactly where, instead of asking Claude to regenerate the whole piece. Level 2: Context
- Ask in chatname the exact spot
- Edit directlyin template artifacts
- Edit with Claudehighlight text in docs, Markdown
- Batch edit requestsacross several Markdown files
-
How Do You Share a Claude Artifact With Other People? Open the artifact, click Share, choose who can access it, and copy the link. Level 2: ContextPro, Max, UltraShare button
- Only you
- Anyone with the link
- Invite by email (beta)
Team, EnterpriseShare button- Only people invited
- Anyone at your organization
- Anyone with the link, if an Owner allows it
-
Can Claude Turn a Rough Idea Into Something You Can Click? Claude can build a clickable prototype as an artifact if you describe the idea and ask for sample data that's labeled as such. Level 2: Context
- Written descriptionmissing steps stay hidden
- Clickable, labeled sample datatest the idea here
- Stored data, connected appslater, adds setup and risk
- A real build from a speconce the idea holds
Projects
Decide how many projects you need, what goes in their instructions and knowledge, and how to keep those files current.
4 techniques →-
How Many Claude Projects Should You Have? Make one project for each context with its own audience, rules and source files, such as one per client or one per product. Level 2: ContextToo few projectsToo many projects
- One for everything
- One per client or product
- One per task
-
How Do You Keep Claude Project Knowledge Up to Date? Remove what's out of date, replace old versions instead of adding new ones beside them, and audit the project on a schedule. Level 2: Context
- 1Find files that disagree
- 2Remove the old version
- 3Check project memory
- 4Ask for a fact and source
Repeat monthly, or when a source changes -
What Should Go in a Claude Project's Instructions? Put in what you'd otherwise retype at the start of every chat: who you are, who the work is for, the tone, the format you want and the things to avoid. Level 2: ContextProject instructions
- Who you are
- Who the work is for
- Tone
- Format you want
- Things to avoid
-
What Files Should You Add to Claude Project Knowledge? Add the documents your answers should be based on: the current brief, the style guide, the price list, the policy, the spec. Level 3: Workflows
- 1Upload files, 30 MB each
- 2All loaded while they fit
- 3Near the limit, RAG turns on
- 4Claude searches for relevant parts
MCP
Add connectors, which are built on MCP, choose which to start with, limit what they can do, and build your own server when none fits.
4 techniques →-
How Do You Build Your Own MCP Server for Claude? Use an official MCP SDK to write a small program that exposes a few tools, test it with the MCP Inspector, then connect it to Claude Code or Claude Desktop. Level 4: Multi-tool systems
- 1Pick 2 or 3 real tasks
- 2Write tools with an SDK
- 3Test with MCP Inspector
- 4Connect, then run evals
-
How Do You Connect Claude to Your Tools? In the Claude apps, add a connector. Level 4: Multi-tool systems
- 1Claude
- 2Connector (MCP server)
- 3Your tool or data
-
How Do You Limit What Claude Can Do Through MCP? Turn off the tools you don't need, approve write actions one at a time, limit scopes when you sign in, and avoid shared admin keys. Level 4: Multi-tool systemsBroad accesslarge blast radius
- shared admin key
- all tools on
- "Allow always" for everything
Least privilegeeasy to review- your own sign-in
- write tools off
- approve each action
-
Which Connector or MCP Server Should You Add First? Add the one service you already copy from most often, and start with read-only use. Level 4: Multi-tool systems
- 1Pick what you copy most
- 2Connect it, read-only
- 3Check results, one or two weeks
- 4Keep it, or remove it
Extended thinking
Extended thinking is now a legacy feature, so learn how adaptive thinking and effort work on hard problems and what the Thinking summary shows.
4 techniques →-
How Do You Get Claude to Work Through Math and Logic? Give Claude the whole problem, raise effort, and ask it to check its result against the numbers you gave. Level 3: WorkflowsDo
- State the full problem
- Raise effort
- Ask for a check against your numbers
Don't- Ask for a one-line answer
- Tell it to answer immediately
- Size max_tokens for the answer alone
-
How Do You See Claude's Thinking, and Can You Trust It? In the Claude apps, expand the Thinking section above a response. Level 3: Workflows
- 1Full thinkingbilled, never shown
- 2Summary you can expandcheck assumptions here
- 3The answerstill needs its own check
-
How Do You Tell Claude What a Right Answer Looks Like? List the conditions a correct answer must meet, in terms that can pass or fail, and ask Claude to verify against them before it finishes. Level 3: WorkflowsNothing to testRuns and reports
- "Make it good"
- Pass or fail conditions
- A test that runs
-
When Should You Turn Up Claude's Thinking? On current Claude models, you don't decide whether Claude thinks. Level 3: Workflows
- lowquick work you can check
- mediumdefault on Opus 5.5, Haiku 5.5
- highdefault on most other models
- xhighlong agentic and coding work
- maxhighest capability
Do you know Claude's less obvious features?
Six questions on artifacts, effort, connectors and projects. Each one explains itself as soon as you answer, and you won't need an account.
What each feature is for
Artifacts are for things you want to keep and use, opened beside the chat. Projects are for work you return to, with standing instructions and files. Connectors, built on MCP, are for live data and actions in other services. Thinking is what Claude does before answering, and effort is how you control it.
Names have changed
If you've followed older guides, a few things won't match. Extended thinking, with its fixed token budget, is now a legacy feature, and current models use adaptive thinking controlled by effort. The Publish button belongs to older artifacts, and current ones use Share. And Claude's interface says connectors where developers say MCP servers.
What are Claude Artifacts?
Artifacts are standalone pieces of content that Claude creates and opens beside the chat, such as documents, slide decks, small tools and dashboards. You can edit, export and share them. Creating them works on all plans, including Free.
What is the difference between Claude Projects and Artifacts?
A project is a workspace with its own instructions, files and chats, used to keep context across conversations. An artifact is a single thing Claude makes, such as a document or tool. Artifacts can be created inside or outside a project.
Does Claude still have extended thinking?
The name survives in places, but on current models thinking is adaptive, meaning Claude decides on each request whether and how much to think. You control it with effort. On Opus 5.5 and Fable 5.1, thinking is always on, and manual token budgets are rejected on Claude 4.7 and later.
What is the difference between MCP and connectors?
MCP, the Model Context Protocol, is the open standard. A connector is what the Claude apps call a connection to a service, and custom connectors use remote MCP servers. In Claude Code and on the API, you'll see the term MCP more often.