Corgi (YC) Got Accused of 'Vibe Coding' a Clone of Papermark's Open-Source Code. Can You Fork an Open-Source Competitor for YC F26?

This weekend, one of the loudest threads on Hacker News was Papermark founder Marc Seitz publicly accusing Corgi, a YC-backed startup last valued around $1.3B, of building a "Dataroom" product by copying Papermark's open-source code and passing it off as "vibe coding." Papermark's core is AGPL-licensed. Seitz's post was blunt: "You didn't vibe code your data room, you stole it from Papermark's open source and enterprise-licensed code... It makes the rest of your business questionable and the YC community look terrible."

Corgi denies it. The company told TechCrunch "No code was used from Papermark," said the offending pages were vibe-coded (AI-generated, not copied line by line), updated them, and sent Seitz a cease-and-desist. Whichever way the audit lands, the episode is a near-perfect case study for founders.

If you're sitting on a YC Fall 2026 application, this fight is more than founder drama. It hits a question almost every technical founder quietly asks an AI model before they apply: can I just fork an open-source product and turn it into my startup? Here's the honest answer.

Can you build a YC startup by cloning an open-source competitor?

Yes, you can build on open source. No, you cannot copy-paste someone's product, strip the license, and call it your own. Those are two completely different things, and the Papermark fight is what the second one looks like when it goes public.

Building on open source is how a huge share of YC companies start. You use Postgres, you fork a library, you stand on a permissively licensed component and add the 90% that's actually your product. That's normal and YC funds it constantly. What Corgi was accused of is different: lifting a competitor's full application, including pages and copy, then presenting it as original work. The HN thread caught it fast because the screenshots showed near-verbatim pages.

The distinction YC partners care about is not "did you use existing code." It's "what did you uniquely build, and would the original author recognize their own work in your product." If the answer to the second part is yes, you don't have a startup. You have a license problem with a landing page.

What does the AGPL actually require?

This is the part most first-time founders get wrong, and it's worth understanding before you build on anything with "GPL" in the license.

Papermark is dual-licensed: its core ships under the AGPL (Affero General Public License) v3, with enterprise directories under a separate commercial license. The AGPL is the strictest common copyleft license. Its defining feature, in section 13: if users interact with your software over a network (i.e. a normal web app or SaaS), you must offer your complete corresponding source code, under the same AGPL, to those users. Permissive licenses like MIT or Apache 2.0 only ask for attribution. The AGPL asks for your whole codebase back.

So if you fork an AGPL project and ship it as a closed-source SaaS, you are out of compliance the moment your first user logs in. You either open-source everything you built on top, or you're infringing. There's no quiet middle path, which is exactly why the HN commenters zeroed in on the license rather than the "vibe coding" framing.

Practical takeaway for your application: know the license of every dependency that touches your core product. MIT and Apache are safe to build a closed product on. AGPL, GPL, and SSPL mean your business model and your license have to agree with each other. If a partner asks "what's your stack" and you can't answer the licensing question, that's a credibility hit in a 10-minute interview.

Will YC reject you for building a clone?

YC won't reject you for competing in a crowded space. They fund direct competitors in the same batch all the time. What sinks a clone application is the absence of an answer to one question: why you, and why is your version going to win?

A clone with no differentiation reads as "I found a product, I copied it, I have no original insight into the market." That's the opposite of what gets funded. Look at the companies that entered saturated categories and still got in: they had a wedge. They were faster, aimed at a specific underserved user, or had a distribution channel the incumbent couldn't reach. "Same product, fewer scruples" is not a wedge.

And there's a reputational layer the Papermark thread makes explicit. Seitz's line that the episode makes "the YC community look terrible" is a reminder that YC is a network with a long memory. The batch is small, founders talk, and "the person who ripped off another founder's repo" is not a reputation you recover from inside that network. YC is making a multi-year bet on you, not just your demo.

How to use open source the right way in your F26 application

If open source is part of your story, frame it as leverage, not as the product:

• Be explicit about what you built versus what you adopted. "We use [permissively licensed component] for X so we could spend our time on Y" is a strength. It shows judgment about where to spend engineering effort.
• Make your differentiation concrete. Name the specific user, workflow, or wedge where you're 10x better. Vague "better UX" claims read as cover for "we copied it."
• Get your licensing clean before you apply. If your core depends on AGPL or GPL code, decide now whether you're open-sourcing or replacing it. Don't let a diligence question surface it for you.
• Have an original insight. The thing the incumbent doesn't believe, or can't act on. That's what partners are actually buying.

The real lesson from the Papermark fight

The blurry part of this whole episode is that AI coding tools make "I wrote this from scratch" and "I cloned this" look identical in a commit history. You can point an agent at a repo and have it reproduce a product in an afternoon. That's exactly why originality of insight, not originality of code, is becoming the thing that separates a fundable founder from a license dispute waiting to happen. Anyone can reproduce the software now. Far fewer people know which software is worth building and why it'll win.

This is the kind of weakness a real YC alum spots in thirty seconds: a slick clone with no defensible reason to exist. Before you submit your F26 application, it's worth having a founder who's actually been through YC pressure-test exactly that, the "why you" question, the differentiation, the parts of your story that sound copied. That's what we built YC Roaster for: brutally honest feedback from people who've sat on the other side of the table. Better to hear "this reads like a clone" from an alum now than from a partner in your interview.

Build on open source all you want. Just make sure the part that matters, the insight, is unmistakably yours.