The way I work with AI has quietly become part of my daily routine. I brainstorm ideas, debug code, draft documents, and research unfamiliar topics, and more often than not, the most valuable output of that work lives inside a single ChatGPT conversation. Some of those threads run for hours. They hold the reasoning behind a decision, the exact snippet that finally fixed a bug, or a carefully refined draft that took a dozen back-and-forth messages to get right. The problem starts the moment I want to keep that conversation, hand it to someone else, or attach it to something more permanent. A long, carefully built thread is genuinely useful, but it is also surprisingly hard to take with me once I close the tab.
\nI kept running into the same small frustration. I would finish a great session, want to save it properly, and end up taking a pile of screenshots or copying raw text into a document where all the formatting fell apart. Code blocks lost their structure, tables collapsed into messy lines, and the clean rhythm of the conversation disappeared. There had to be a better way to turn a living chat into something I could actually file away and reuse.
\nThis is where Save ChatGPT as PDF steps in, an open-source project that turns any public ChatGPT share link into a clean, faithfully formatted PDF in a single click. Just paste a link in the form of https://chatgpt.com/share/<uuid>, and the entire conversation, including markdown, code blocks, tables, math, and ChatGPT's familiar turn-by-turn layout, is rendered into a document you can download, archive, and share anywhere. No sign-up, no browser extension, and no fiddling with print settings. You can try it live at save-chatgpt-as-pdf.membasuh.com.
ChatGPT already offers a built-in Share Link feature, and for quick, casual sharing it works well enough. You generate a link, send it to a friend, and they open it in their browser. For a throwaway conversation, that is perfectly fine. But the moment a conversation needs to travel beyond a browser tab, or needs to last longer than a passing moment, the limitations of a share link become obvious:
\nA PDF solves all of this in one step. A PDF file is universally accessible, permanently shareable, printable, and documentable. It opens on any device, survives long after the original link is gone, and slots naturally into the places where real work is stored and reviewed. It can be searched, highlighted, annotated, signed, and dropped into a folder alongside everything else that matters. Think about attaching an AI conversation to a job application as proof of how you approached a take-home assignment, including a research thread in a thesis appendix, dropping a troubleshooting session into a client report so the whole team can follow the reasoning, or embedding a planning discussion into a Notion page that lives next to the rest of your project notes. In every one of those cases, a share link is a fragile dependency, while a PDF is a self-contained artifact that you fully own.
\nSave ChatGPT as PDF treats the export not as a screenshot or a rough dump, but as a proper document. That shift, from a temporary link to a durable file, is exactly the upgrade that makes an AI conversation suitable for professional, academic, and archival use. The conversation stops being something locked behind a login and becomes something you can keep forever, exactly as it looked when it mattered.
\nAt its core, the project is built on Next.js 15, React 19, and TypeScript, with a clean interface styled using Tailwind v4 and shadcn/ui. From the outside, the flow could not be simpler. You paste a public share link, the app fetches and reads the conversation, and a few moments later a polished PDF is ready to download. Underneath that simplicity, though, there is a deliberate pipeline designed to keep the output faithful and the service safe.
\nOnce the conversation data is parsed, it moves into the rendering stage. The document is produced with @react-pdf/renderer, which gives precise, code-level control over typography, spacing, page breaks, and layout. For anything that needs richer handling, pdfjs-dist comes into play, helping ensure the final file behaves like a true PDF rather than a flattened image. The renderer rebuilds the familiar ChatGPT reading experience on the page: user turns and assistant turns are visually distinct, code keeps its monospaced styling and syntax highlighting, tables keep their columns, and math renders as math rather than broken symbols. The goal throughout is fidelity, so that opening the PDF feels like reading the original conversation, just in a format you own.
To keep the experience fast, the project uses an optional Upstash Redis cache. The first time a particular conversation is rendered, the result can be stored, so any repeat visit to the same link returns almost instantly instead of running through the whole pipeline again. Because it is optional, you can run the app entirely without it and everything still works, which keeps local setup and self-hosting friendly.
\nFinally, the whole thing is built to run at the edge. It deploys to Cloudflare Workers via OpenNext, which places the app close to users around the world and keeps it lightweight and scalable. The entire experience is designed to be effortless from the user's side: no sign-up, no configuration, just a link and a download.
\nA tool that accepts arbitrary URLs and turns them into documents has to be careful, and this is an area I genuinely cared about while shaping the project. Two principles guide it: take in as little as possible, and keep what comes in from causing harm.
\nOn the data side, the approach is simple. There is no sign-up and no persisted user data. You do not create an account, you do not hand over personal details, and your conversations are not quietly stored to build a profile of you. The app does the job you asked for and then gets out of the way. The optional cache exists purely to speed up repeat rendering, not to collect anything about who you are.
\nOn the safety side, the app includes three layers of SSRF protection. Server-Side Request Forgery is a class of attack where a malicious user tricks a server into making requests it should not, for example reaching into internal networks or private addresses. Because this tool deliberately fetches a remote URL on the server, that risk has to be taken seriously, and the layered checks are there to make sure only legitimate public ChatGPT share links are ever fetched. On top of that, rate limiting keeps any single source from overwhelming the service, so it stays responsive and fair for everyone. None of this is visible to a normal user, and that is exactly the point: the safe path is the default path.
\nThe project is built using the following tools and frameworks:
\nThe value of a clean, permanent export becomes clearest when you picture the moments people actually reach for it. A few that come up again and again:
\nThe fastest way to try it is simply to visit save-chatgpt-as-pdf.membasuh.com, paste a public ChatGPT share link, and download your PDF. There is genuinely nothing else to it for everyday use. If you would rather run it yourself, study the code, or self-host your own instance, getting started locally is straightforward:
\nClone the Repository
\nStart by cloning the project from GitHub:
\ngit clone git@github.com:fyvri/save-chatgpt-as-pdf.git && cd save-chatgpt-as-pdfInstall Dependencies
\nInstall everything the project needs:
\nnpm installSet Up Your Environment
\nCopy the example environment file and adjust it to your needs. The Upstash Redis cache is optional, so you can leave those values empty to run without caching while you are getting started.
\nRun the Application
\nLaunch the development server locally:
\nnpm run devThen open http://localhost:3000, paste a public ChatGPT share link such as https://chatgpt.com/share/<uuid>, and watch the conversation turn into a polished PDF. When you are ready to put it online, the project is set up to deploy to Cloudflare Workers through OpenNext, so your own edge-hosted instance is only a few steps further.
Save ChatGPT as PDF is more than a quick utility. It is a small step toward making AI conversations something you truly own: durable, portable, and ready for serious use. Whether you are a student archiving research, a professional documenting work, or a developer who wants to explore and extend the project, there is something here for you. The project is open source, and that openness is an invitation. Ideas, bug reports, and improvements are all welcome, and every contribution helps make the tool better for the next person who needs it.
\nDive into the GitHub repository, try it out at save-chatgpt-as-pdf.membasuh.com, and feel free to open an issue or a pull request.
\nWe are generating more valuable conversations with AI than ever before, yet most of them quietly vanish the moment we close the tab. Save ChatGPT as PDF exists to fix that small but real problem, turning a fleeting share link into a clean document you can keep, share, and rely on. It started as a personal itch, the simple wish to hold on to the good conversations, and grew into a tool I hope makes that just as easy for you. If even one important chat ends up safely saved because of it, the project has done its job. Together, let's make every great conversation worth keeping.
","excerpt":"Overview The way I work with AI has quietly become part of my daily routine. I brainstorm ideas, debug code, draft documents, and research…","tableOfContents":"