You've got a codebase. Maybe dozens of .cpp files. You want Claude to help you generate documentation. Simple enough, right?

You drag and drop your files, hit send, and then... nothing.

"This isn't working right now. You can try again later."

Sound familiar? Here's what's actually happening when Claude hits the wall on large file uploads.

You're Probably Hitting the Context Limit

Claude can only process so much text at once—this is called the "context window." When you upload a stack of C++ files, each one eats into that limit. Headers, implementations, comments, boilerplate... it adds up fast.

Once you exceed what Claude can hold in memory, it doesn't gracefully summarize or ask you to trim things down. It just fails. And you get the message.

File Processing Takes Serious Resources

Parsing multiple code files isn't lightweight work. Claude has to read, understand structure, track relationships between classes and functions, and then generate coherent documentation. When you throw a lot of files at it simultaneously, the backend can run out of processing headroom—especially during busy hours.

The Error Message Tells You Nothing Useful

Here's the frustrating part: whether you've exceeded token limits, overwhelmed the server, or triggered some internal timeout, you get the exact same vague message. There's no "Hey, that's too many files" or "Try uploading in smaller batches."

Just: "This isn't working right now."

Thanks, Claude. Very helpful.

What Actually Works

If you're trying to document a large codebase, here's how to avoid the error:

• Upload in smaller batches. Start with 3–5 related files at a time instead of the whole project.
• Prioritize key files. Feed Claude the core logic first—main modules, important classes—then expand.
• Ask for incremental output. Generate docs for one component, then move to the next in a follow-up message.
• Trim the fat. Remove autogenerated code, redundant headers, or third-party libraries before uploading.

The Bottom Line

Claude is powerful, but it has real limits—especially with bulk file processing. The generic error message makes it feel random, but it's usually a capacity issue in disguise.

Until Anthropic gives users clearer feedback (like "You've uploaded too much—try fewer files"), the workaround is simple: go smaller, go slower, and don't expect Claude to swallow your entire repo in one bite.