Give AI the right context
from your codebase

repoprep is a browser-based tool that bundles your entire codebase — or any collection of files — into a single structured text file that AI models can actually read. Instead of pasting files one by one into Claude, ChatGPT, or Gemini, you upload your project once and get a clean, AI-ready context file in seconds.

Never. All processing happens entirely inside your browser using Web APIs. Your files are read locally, processed locally, and the output is generated locally. Nothing leaves your machine. repoprep has no backend, no database, and zero telemetry — your code stays yours. The parsing engine that does this is open source, so you don't have to take our word for it — you can read the exact code at github.com/Kishoren1/repoprep-app.

Any AI that accepts text input: Claude (Anthropic), ChatGPT (OpenAI), Gemini (Google), Cursor, GitHub Copilot, Mistral, and any other model with a context window. The output is plain text with a project tree header followed by each file's content — a format all major models handle well.

Yes — the repoprep Chrome extension lets you generate a context file directly from any GitHub or GitLab repo page, without downloading the repo first. It also supports local folders and ZIP files, the same as the web app. Pro is a single licence — verify with the same email and license key on both the extension and the web app to unlock it on each, at no extra cost. Available free on the Chrome Web Store.

repoprep reads virtually every developer file type: JavaScript, TypeScript, Python, Go, Rust, Ruby, Java, C/C++, Markdown, JSON, YAML, TOML, HTML, CSS, SQL, shell scripts, and more. It also extracts text from PDF and DOCX files. Binary files like images are automatically skipped.

The free plan supports up to 50 files and 4 MB of total input — plenty for small components, bug reports, or quick questions. Pro ($6 one-time, no subscription) raises those limits to 250 files and 20 MB, covering full-scale monorepos and large projects. You pay once and own it permanently.

Once repoprep generates your context file, click 'Copy to clipboard' and paste it directly into Claude, ChatGPT, or any other AI chat. For models that support file uploads (like Claude Projects or the ChatGPT file tool), you can also download the .txt file and attach it. Either way, the AI instantly has full awareness of your entire project.

Yes — repoprep automatically skips common noise directories like node_modules, .git, dist, build, .next, __pycache__, and others. You get only the source files that matter. The Pro plan handles projects up to 20 MB and 250 files, covering most real-world monorepos.

The token estimator uses a character-based heuristic (approximately 4 characters per token) that is accurate to within about 10–15% for typical source code. It is intentionally shown as an estimate, not a guarantee — context window limits vary by model and version, so treat it as a guide rather than a hard number.

Because repoprep processes everything inside your browser tab rather than on a server, the tab itself has to do the work a backend normally would. An unlimited cap would eventually let someone load a project large enough to freeze or crash that tab. 250 files and 20 MB is set comfortably above what most real-world projects need, while staying within what a browser can reliably handle without slowing down or crashing.