Use Claude.ai Project knowledge for reusable source material, Project instructions for guidance that should shape every chat in that workspace, and Claude Code’s CLAUDE.md files for local project or user memory in a coding environment. Skills are a separate concept, but the official information available here does not establish their current format or how Claude discovers and invokes them, so this comparison does not assign them setup rules.
First, separate Claude.ai Projects from Claude Code
“Reference files” can mean different things depending on which Claude product you use. In Claude.ai, they usually mean uploaded material in a Project’s knowledge base. In Claude Code, context and guidance can live in files such as CLAUDE.md in or around a working directory. These are different workflows, not two interfaces for the same feature.
Anthropic describes Claude.ai Projects as self-contained workspaces with their own chat histories and knowledge bases. Project instructions further tailor Claude’s responses in that workspace. See Anthropic’s overview of Projects and guide to creating and managing them.
What each option is for
| Option | Where it lives | Main purpose | Scope |
|---|---|---|---|
| Claude.ai Project knowledge | Uploaded reference material in a Project | Provides source material or background Claude can draw on | Chats within that Project |
| Claude.ai Project instructions | Project settings in Claude.ai | Guides how Claude should behave and respond in that workspace | Chats within that Project |
| Claude.ai profile preferences and styles | Claude.ai personalization settings | Profile preferences apply more broadly; styles control tone and format | Account-wide preferences or response presentation, as applicable |
Claude Code CLAUDE.md |
Local files associated with a coding environment | Stores project or user memory and guidance; a file may contain both context and instructions | Depends on the file’s location and Claude Code’s documented memory behavior |
| Skills | Not established by the official materials cited here | Separate concept; current conventions are not established here | Not established here |
Anthropic says Project knowledge is used in all chats within its Project, while chat context is not shared across Project chats unless information is added to Project knowledge. Project instructions also apply to chats in that Project. Anthropic’s personalization guide distinguishes those project-level controls from profile preferences and styles.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsChoose Project knowledge for reusable facts and source material
Add documents to Project knowledge when Claude should be able to consult them across conversations in one focused workspace—for example, product documentation for a support Project or reference material for a writing Project. Treat the files as inputs to draw on, rather than as the place to put every behavioral preference.
Anthropic’s upload help page lists PDF, DOCX, CSV, TXT, HTML, ODT, RTF, EPUB, JSON, and XLSX among supported formats; XLSX requires the analysis tool to be enabled. The page states a 30 MB per-file limit and says the number of files in Project knowledge is unlimited, subject to the total content fitting available context capacity. Check the current upload guidance and your plan’s availability before relying on a particular limit.
For large collections, Anthropic says retrieval-augmented generation (RAG) activates automatically near context limits and can expand Project knowledge capacity by up to 10x. That is Anthropic’s product description, not an independent performance measurement; the RAG help page does not establish how accurately a particular document will be retrieved for a particular question.
Choose Project instructions for workspace-wide behavior
Put directions in Project instructions when they should influence how Claude responds across chats in that Project: for example, the audience to write for, the format to use, or constraints to follow. Anthropic’s Help Center states, “Claude will use these instructions for all the chats within the project.” Do not confuse that scope with account-wide preferences or a source document that Claude should consult.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Use Claude Code memory files for coding-environment context
Claude Code documentation describes file-based memory with distinct project and user scopes, including ./CLAUDE.md for project memory and ~/.claude/CLAUDE.md for user memory. It also documents local project memory, imports, and recursive file discovery. The files load at startup, according to Anthropic’s memory documentation.
This is a filesystem workflow, not Claude.ai’s Project knowledge UI. A repository can carry project-specific guidance alongside code, while user memory represents a different scope. Do not assume a file is shared with teammates or synchronized across accounts simply because it exists in a repository; sharing and behavior depend on how the repository and each user’s setup are configured. Anthropic’s cited Claude Code memory documentation is in Chinese, so consult the current official documentation for exact setup details rather than treating Claude.ai Project behavior as a substitute.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Where Skills fit—and what is not established
Skills belong in the comparison as a separate concept, not as another name for Project knowledge or CLAUDE.md. The official materials cited here do not establish current Skills conventions for file format, discovery, or invocation. That means the documented startup, import, and discovery behavior of Claude Code memory files should not be attributed to Skills, and this article does not offer Skills setup steps.
Quick Recap
Best Value
A practical placement decision
- Claude.ai, “What facts or documents should Claude consult in this workspace?” Put the material in that Project’s knowledge base.
- Claude.ai, “How should Claude respond in every chat in this workspace?” Put the guidance in Project instructions.
- Claude.ai, “Should this preference apply beyond one Project?” Review profile preferences; use styles when tone or response format is the main concern.
- Claude Code, “What context or guidance should accompany this coding environment?” Use the documented
CLAUDE.mdmemory workflow, choosing the project or user scope that matches the need. - “Should I put this in a Skill?” Verify the current official Skills documentation first; the cited material does not support specific Skills conventions.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Free tools Windows power users keep installed
One-click scans. No signup required.
