How to feed multiple files to Claude at once

Claude's file handling

Claude supports file uploads through the chat interface. You can attach PDFs, text files, code, images, and other document types. Claude processes each file and includes its content in the conversation context.

The challenge comes when you have a project with many files. If you need Claude to see a research folder with 15 PDFs, or a codebase with dozens of source files, uploading them individually is slow and does not guarantee the AI will synthesize the content across all files effectively.

The practical approach: merge before uploading

The most reliable way to give Claude comprehensive project context is to merge your files into one structured text file before uploading. This gives Claude a single, coherent document to work with rather than a set of disconnected attachments. The content is organized with clear section headers, making it easy for Claude to navigate and cross-reference.

Using Claude Projects for persistent context

Claude Projects is Anthropic's built-in solution for persistent file context. You can upload files to a Project, and they remain available across all conversations within that project. This is useful when you work on the same project over multiple sessions.

The limitation is that Claude Projects only works within Claude. If you also use ChatGPT, Gemini, or local models, your project context is locked to one platform. A model-agnostic approach, where you build a reusable bundle that works in any AI tool, gives you more flexibility.

How much fits in Claude's context window

Claude's 200,000-token context window holds roughly 150,000 words, which is the equivalent of a 500-page book. A typical project bundle of 10 to 20 well-curated documents (without images or binary files) usually fits comfortably.

If your bundle is too large, prioritize the most relevant files. Include your project brief, key reference documents, and recent work. Leave out archived material, daily logs, and anything the AI does not need for the current task. Riflet shows a live token count so you can see exactly how much space your bundle occupies before exporting.

Best format for Claude uploads

Markdown works best. Claude handles Markdown natively and uses the structure (headers, code blocks, lists) to navigate the content efficiently. When your files are merged into a Markdown bundle with clear section headers, Claude can locate specific parts of your context when answering detailed questions.

Plain text also works. The key is that the content is structured and labeled so the AI knows where one source ends and another begins.