Most of the useful things in Claude are not hidden, but they sit in menus most people never open: Projects, memory settings, incognito chats, Artifacts and Skills. This guide lists twelve of them with the exact menu path for each, based on Anthropic's own help-center and documentation pages as checked on 2026-11-17. Anthropic changes its layouts often, so where a label has more than one name the article says so.
Who this is for and what you will have
This guide is for anyone who uses Claude at work or at home and suspects they are using a fraction of it: office staff, small business owners, IT admins and developers. At the end you will have a short list of settings to review, a few habits that improve answers, and a clear idea of which features need a paid plan.
The tips are grouped into four parts: organizing your work (tips 1 to 3), controlling what Claude remembers (4 to 6), making things (7 to 10) and writing better prompts (11 and 12). If you want ready-made prompts rather than features, read 25 Claude Prompts for Work You Can Copy and Paste.
Before you start: plans and the file-creation setting
Several features below depend on a single setting. Anthropic's Artifacts article says to turn on Create and edit files in Settings > Capabilities. The same toggle is labeled Cloud code execution and file creation in older layouts. On Team and Enterprise plans, an admin turns it on under Organization settings > Capabilities. Skills depend on the same code-execution capability, so if a feature seems missing, check this toggle first.
Plan availability matters too. In the sources checked:
| Feature | Availability (per Anthropic help center) |
|---|---|
| Projects | All users, including free (up to five projects on free) |
| Sharing a project | Team and Enterprise |
| Artifacts in chat, plus Docs, Slides and Design templates | All plans |
| Data, Sheets (beta) and Whiteboards (beta) templates | Pro and higher |
| Skills | Free, Pro, Max, Ultra, Team and Enterprise (requires code execution) |
| Search past chats | Paid plans only |
| Incognito chats | All plans |
Part 1: Organize your work
Tip 1: Use a Project instead of starting from scratch each time
Projects are workspaces where you group chats and give Claude shared context. Anthropic's article says project knowledge and project instructions apply to every chat in the project, and that context does not carry across chats unless it is in the project knowledge base.
To create one, open Projects from the left side of your account or go to claude.ai/projects, then click + New Project in the upper right corner and enter a name and description. The help article notes that Claude cannot see the name and description, so put anything Claude needs to know in the instructions instead.
Tip 2: Set project instructions once
Inside a project, click Set project instructions, write how Claude should behave, and click Save instructions. Claude applies them to all chats in that project. Good instructions are short and specific: who the audience is, the tone, the output format and what to avoid.
You help the support team at Contoso write customer replies.
Audience: small business owners, not technical.
Format: a greeting, 2 to 4 short paragraphs, a clear next step.
Tone: polite and direct. Do not promise refund dates.
If the question is missing information, ask for it before answering.Adjust the wording to your own team. For a comparison with ChatGPT's equivalent, see ChatGPT Custom GPTs vs Claude Projects for team knowledge.
Tip 3: Add knowledge files, and know their limits
On the project's main page, click the + button to upload documents, text files or code snippets. Claude uses them as context in the project's chats. Anthropic's upload article gives the limit for project files as 30MB per file, with no fixed cap on the number of files, but the total content must fit within Claude's context window. The Projects article adds that on paid plans, when project knowledge nears the context window limit, Claude automatically enables retrieval-augmented generation (RAG) mode to expand capacity.
If you work in a team, the Share project button (Team and Enterprise only) lets you choose Can view or Can edit for each member, so one person can maintain the instructions while others only chat.
Part 2: Control what Claude remembers
Tip 4: Know where memory lives and turn it off if you want
Per Anthropic's memory article, memory is on by default for Free, Pro, Max and Ultra plans on web, desktop and mobile. On Team and Enterprise, owners control availability. To review it, go to Settings > Your Claude (or Settings > Memory in the older layout) and select Manage next to Your memory. Memories are listed under Topics; select one to read it, then use the edit icon or choose Delete. You can also tell Claude in a chat what to remember, change or forget.
Turning the Generate memory from chats toggle off gives you two options: Pause memory keeps existing memories, while Reset memory permanently deletes them. Choose carefully, because reset cannot be undone. The article also lists Include sensitive topics in memory, which is off by default.
Tip 5: Turn memory off for a single chat, or use incognito
You do not have to switch memory off globally. On web and desktop, start a new chat, click +, click Memory, then send your first message. On mobile, tap +, turn off the Memory switch, close the sheet and send your first message. The article stresses you must do this before sending the first message.
For a chat that should leave no trace, start an incognito chat: open a new chat outside a project and click the ghost icon in the upper right corner. Incognito chats are available on all plans. Claude will not remember them, they are not saved to memory or chat history, and Claude will not pull from them when searching previous conversations. On Team and Enterprise, the article notes that incognito chats are still included in standard data exports and follow the organization's retention policies, so incognito is not the same as invisible to your employer.
Tip 6: Search past chats, or disable it
On paid plans, Claude can search your earlier conversations. Searches outside projects and inside each project are separate, and each search appears in the chat as a tool call. The toggle is labeled Search and reference chats under Settings > Your Claude (Settings > Memory in older layouts); switch it off to stop Claude searching past chats. The article also says this feature does not work for Enterprise organizations using customer-managed encryption keys.
A practical use: instead of pasting last week's notes again, ask Claude to find the earlier conversation about a topic and continue from it.
Part 3: Make things, not just answers
Tip 7: Ask for an Artifact
Anthropic describes Artifacts as self-contained pieces of work such as a document, design, deck, dashboard or small interactive tool. They open beside your conversation, where you can edit, revisit and share them, and saved ones appear in the Artifacts tab in the sidebar. Ask for one in any chat, or start from a template by selecting Output in the message box or choosing one in the Artifacts tab.
Two small controls are easy to miss. In a doc, highlight text and click Edit with Claude to request a change to just that section. If an artifact shows an error, click Try fixing with Claude; the help article says success is not guaranteed.
One change to be aware of: the same article says legacy artifacts created in chat before September 16, 2026 can still be published and shared, but you cannot create new ones. New artifacts start private, and you use the sharing controls to decide who can open them.
Tip 8: Export instead of copy and paste
Template-based work can be exported from Export. Per the article, docs can go to Word, PDF, Markdown or Google Docs; decks to PowerPoint, PDF or Google Slides; designs to .zip, PDF, PowerPoint or HTML; and animations export as MP4. For a weekly report, this means you can draft in Claude and hand over a Word file without reformatting.
Tip 9: Use built-in Skills for Office files
Skills give Claude specialized knowledge and workflows. Anthropic provides built-in skills for Excel spreadsheets, Word documents, PowerPoint presentations and PDFs, and with code execution and file creation on, Claude uses them automatically when relevant. To manage them, go to Customize > Skills and toggle skills on or off. On Team and Enterprise, an owner first goes to Organization settings > Plugins & skills, opens the Policy tab and enables the file-creation setting and Skills.
To add a custom skill, package the skill folder as a ZIP, go to Customize > Skills, click +, then + Create skill, choose Upload a skill and upload the ZIP. Uploads can fail if the ZIP is too large, the folder name does not match the skill name, the skill.md file is missing, or the name or description has invalid characters.
Safety comes first with third-party skills. Anthropic's article calls prompt injection and data exfiltration the most significant risks, says skills may include or instruct Claude to install third-party packages, and advises installing skills only from trusted sources. For anything shared by a colleague or found online, read the bundled files, especially scripts, dependencies and instructions to connect to external networks, before enabling.
Tip 10: Check upload limits before you fight an error
From Anthropic's upload article, chat uploads allow 500MB per file and 20 files per chat, with images up to 8000x8000 pixels. PDFs can have up to 1000 pages; larger files show an "Uploaded file is too large" error. PDFs of 100 pages or fewer get analysis of text and visual elements such as charts; PDFs of 101 to 1000 pages are processed as text only. So if a chart-heavy 300-page report gives poor answers about its graphs, split it into parts of 100 pages or fewer.
Supported document types are PDF, DOCX, CSV, TXT, HTML, ODT, RTF, EPUB, JSON and XLSX (XLSX needs code execution and file creation enabled), and images in JPEG, PNG, GIF and WebP.
Part 4: Write prompts the way the documentation recommends
Tip 11: Give examples, and give 3 to 5
Anthropic's prompting best-practices page says examples are among the most reliable ways to steer output format, tone and structure. It recommends making them relevant and diverse, wrapping them in <example> tags (multiple examples in <examples> tags) so Claude can tell them from instructions, and including 3 to 5 for best results.
Rewrite customer complaints as one-line ticket titles.
<examples>
<example>
Input: My invoice for October shows the wrong VAT number again!
Output: Invoice October: wrong VAT number
</example>
<example>
Input: I can't log in on my phone since the update.
Output: Mobile login failing after app update
</example>
<example>
Input: Where is my order, it was meant to come Monday
Output: Order delayed past promised Monday delivery
</example>
</examples>
Now rewrite this one:
Input: <paste complaint here>Tip 12: Put long documents first, then your question
For large inputs (the page says 20k+ tokens), Anthropic advises placing long documents near the top of the prompt, above your query, instructions and examples, and says this improves performance across all models. The same page notes that queries at the end can improve response quality by up to 30 percent in tests, especially with complex multi-document inputs. With several documents, it suggests wrapping each in <document> tags with <source> metadata.
Also run the page's "key test" on any prompt that disappoints: show it to a colleague with minimal context and ask them to follow it. If they would be confused, Claude will be too. Be specific about the output format and constraints, and if you want more than the bare task, ask for it explicitly.
Troubleshooting
- A feature is missing from the menu. Check Settings > Capabilities for Create and edit files first, then your plan in the table above. On Team and Enterprise, an owner may need to enable it under Organization settings.
- "Uploaded file is too large". The PDF has more than 1000 pages or the file exceeds the size limit. Split it.
- The menu names do not match this guide. Anthropic's own pages mention older and newer layouts (for example Settings > Memory versus Settings > Your Claude). Look for the same toggle label.
- A skill upload fails. Check the ZIP size, that the folder name matches the skill name, that skill.md exists, and that the name and description use valid characters.
- Claude seems to ignore earlier context. Context does not carry across chats outside a project unless it is in the project knowledge or memory. Put stable facts in project instructions.
Closing checklist
- Create one Project for your most repeated task and add instructions.
- Review Your memory under Settings > Your Claude and delete anything you do not want kept.
- Use incognito or the per-chat Memory switch for sensitive chats.
- Try one Artifact and export it.
- Read the files of any skill before enabling it.
- Add 3 to 5 examples and put long documents above your question.
What we checked
- Opened Anthropic's help-center pages on Projects, file uploads, Artifacts, Skills and memory on 2026-11-17 and copied the menu labels exactly as written there (for example "+ New Project", "Set project instructions", "Search and reference chats").
- Cross-checked the file-creation setting between the Artifacts page (Create and edit files, formerly Cloud code execution and file creation) and the Skills page (Code execution and file creation, called Create and edit files in the updated layout); both point to Settings > Capabilities.
- Cross-checked the 30MB project-file limit and 500MB chat-file limit against the single upload article; the Projects article gives no fixed knowledge capacity figure, so none is stated here.
- Read the prompting best-practices page on platform.claude.com for the 3 to 5 examples guidance, the long-documents-first advice and the "key test"; the 30 percent figure is quoted as Anthropic's "in tests" claim, not measured here.
- Not checked: a separate help-center page on styles returned errors and search snippets disagreed on whether styles are moving to skills, so styles are deliberately left out.
- Not tested in a live Claude account; steps follow Anthropic's documentation as of 2026-11-17, and the example prompts have not been run.
Last tested: 2026-11-17
References
- Anthropic Help Center, create and manage Projects: https://support.claude.com/en/articles/9519177-how-can-i-create-and-manage-projects
- Anthropic Help Center, uploading files to Claude: https://support.claude.com/en/articles/8241126-uploading-files-to-claude
- Anthropic Help Center, what are Artifacts: https://support.claude.com/en/articles/9487310-what-are-artifacts-and-how-do-i-use-them
- Anthropic Help Center, using Skills in Claude: https://support.claude.com/en/articles/12512180-using-skills-in-claude
- Anthropic Help Center, chat search and memory: https://support.claude.com/en/articles/11817273
- Claude Docs, prompting best practices: https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/claude-prompting-best-practices
Comments
Questions and tips are welcome. Comments are checked by the EpiFive team before they appear; please do not share passwords, phone numbers or other private details.