85K-Star GitHub Project Accused of "Code Laundering": The Hermes Agent Drama and Where Open Source Draws the Line

The biggest tech drama last week was probably the Hermes Agent incident.

On April 15th, Chinese AI team EvoMap published a lengthy post accusing Hermes Agent of plagiarism—their self-evolution and persistent memory features closely matched EvoMap’s work, and the timeline for Hermes’s GitHub repository creation didn’t align with their claimed dates.

Nous Research, Hermes’s developer, responded: “Our repo existed in July 2025. We were first. Delete your account.”

Then… they deleted that response.

My take

The most frustrating part isn’t “who copied whom,” it’s that Nous Research’s response was so poorly executed. If they really had code from July 2025, why not just show the commit history? Deleting the response makes things even more suspicious.

Where exactly does “originality” end in open source?

This is an old question without a clear answer. My understanding:

• Using an open source project’s capabilities to build your own product—that’s fine
• Referencing an open source project’s architecture and implementing your own version—gray area
• Directly copying code and just changing variable names—code laundering, no way to spin it

The Hermes case currently lacks clear evidence. EvoMap claims “high feature overlap,” but how much overlap constitutes plagiarism? This boundary is genuinely hard to quantify in engineering terms.

As someone who participates in open source communities, what concerns me more: will this incident discourage people from open-sourcing their work? If every open source release comes with plagiarism accusations, who will want to share code going forward?

Here’s hoping for a fair conclusion.