What is inside a Project
A Project bundles four things that every chat in it can use:
| Part | What it is | Good for |
|---|---|---|
| Knowledge | Files and text you add once | Style guides, product docs, contracts, code, meeting notes |
| Instructions | Standing guidance for every chat in the Project | Tone, format, role, what to always or never do |
| Chats | All conversations started inside the Project | Keeping one topic’s work together |
| Memory | A memory that belongs to this Project only | What Claude learns while you work on this topic |
How Claude handles a large knowledge base
On every plan, Claude reads Project knowledge into its context. On paid plans, when the knowledge gets close to the context limit, Claude switches to retrieval-augmented generation (RAG): instead of reading everything, it searches the knowledge and pulls in the relevant parts. Anthropic says this gives up to ten times more capacity. Anthropic does not publish an exact size limit.
That switch has a practical consequence. With a small knowledge base Claude sees everything at once. With a large one, it sees what the search finds. A question that needs two facts from opposite ends of a big document set can miss one of them. Keep files focused and named clearly, and put the rules that always apply in the instructions, not buried in a file.
How memory works inside a Project
Each Project keeps its own memory. Inside a Project, Claude combines that Project memory with memory from chats outside any Project, but it never uses the memory of your other Projects. To keep a Project fully separate, for example per client, turn off “Use account memory” on the Project page. See does Claude have memory for the full picture.
Plans and limits
- Free: up to 5 Projects.
- Pro, Max and Ultra: unlimited Projects.
- Team and Enterprise: unlimited Projects, plus sharing with colleagues at view or edit level. See sharing Claude Projects with your team.
Project content is also cached, which means reusing the same knowledge counts less toward your usage limits than pasting it into new chats. The cache expires after a period of inactivity.
Anthropic is rolling out a new beta version of Projects in phases, starting with Claude Code, in which a Project becomes one conversation with parallel threads in the cloud. Chat and Cowork Projects stay on the current version for now.
What Projects are good for, and what not
Use a Project when the same context comes back again and again: a client, a product, a codebase, a recurring report. Do not use one Project for everything: separate Projects keep memory and retrieval focused.
A shared Project can serve as a team knowledge base, but it holds what someone uploaded at some point. Uploaded files go stale when the source documents change unless someone maintains them, and the Project only exists in Claude.
Where Bombos fits
A Project holds what someone uploaded into Claude. Bombos is a plugin that gives every AI tool a team uses, Claude and Claude Code included, one shared memory that reads what already lives in mail, documents and repos and keeps the rules your team has taught it. It runs in the cloud or on your own server, and one person can try it first for their own agents.
Related questions
How many files can a Claude Project hold?
Anthropic does not publish a fixed number. On paid plans, Claude switches to retrieval when the knowledge nears the context limit, which expands capacity up to ten times.
Can I share a Project on Pro?
No. Sharing Projects is only available on Team and Enterprise.
Does a Project remember things from my other Projects?
No. Each Project has its own memory. Inside a Project, Claude also uses memory from chats outside Projects, unless you turn off Use account memory.
Do Projects save usage?
Yes, somewhat. Project content is cached, so reusing it counts less toward your limits than pasting the same files again. The cache expires after inactivity.
