Hermes Agent Accused of Plagiarism—Founder Retorts with “Delete Your Account”

A small open‑source team claims Hermes Agent copied its self‑evolution architecture within 36 days, detailing ten code‑level similarities, while the project's founder responded on Twitter with a terse “Delete your account,” sparking a heated debate over plagiarism and attribution in AI agents.

AI Insight Log
AI Insight Log
AI Insight Log
Hermes Agent Accused of Plagiarism—Founder Retorts with “Delete Your Account”

In the open‑source AI Agent community a controversy erupted when the 2,000‑star EvoMapAI team accused the 8.8‑k‑star Hermes Agent of copying its core self‑evolution architecture, prompting the founder of Nous Research to reply on Twitter with the blunt phrase “Delete your account.”

Hermes Agent, released by Nous Research, quickly became known as the strongest rival to OpenClaw (the “lobster”) thanks to its “self‑evolution” system that lets agents extract experience from tasks and automatically refine skills and prompts.

On February 1, EvoMapAI open‑sourced its Evolver engine, which also implements a self‑evolution mechanism via a Genome Evolution Protocol (GEP) that packages agent experience into “gene capsules” for global sharing. Evolver attracted over 1,800 stars.

The timeline is tight: Evolver was released on February 1, Hermes Agent v0.1.0 appeared on February 25, the independent repository hermes-agent-self-evolution was created on March 9, and Hermes Agent v0.2.0 launched on March 12—only 36 days between Evolver’s public release and Hermes’s self‑evolution repository.

EvoMapAI published a detailed code‑comparison report highlighting three major overlaps: (1) a 1:1 correspondence in the task‑loop and asset‑extraction paradigm, where Hermes’s loop mirrors Evolver’s ten‑step cycle with only terminology changes; (2) an identical three‑layer memory system (Factual, Procedural, Search); and (3) virtually the same periodic reflection and dynamic skill‑loading mechanism, including identical trigger cycles, weighted scoring functions, and cross‑language design patterns such as atomic writes, safety scans, injection guards, and capacity controls. They also documented twelve term replacements and noted that Hermes never cited Evolver in any of its eight public materials.

EvoMapAI’s founder posted on Twitter, “We aren’t here to play judge. We’re just putting the code comparisons on the table. The hard work of indie open‑source creators shouldn’t be erased like this.”

Nous Research responded with a tweet claiming “Hermes Agent is an overnight success nine months in the making,” attaching a GitHub commit graph that shows the main repository was created on July 22 2025. However, that repository was private at the time, so its code could not be inspected, and the self‑evolution module only became visible in the March 9, 2026 independent repo.

In its defense, Hermes’s team pointed out that the self‑evolution module relies on the GEPA (Genetic‑Pareto) framework, an independent academic project from UC Berkeley and Stanford presented at ICLR 2026 (oral presentation) and already integrated into Stanford NLP’s DSPy library, indicating a scholarly origin separate from Evolver.

The community reaction split: one camp argued that a 36‑day window and dozens of near‑identical design choices make coincidence unlikely, especially given the zero citations for Evolver’s concepts. The other camp contended that three‑layer memory, skill loading, and reflection loops are common paradigms in the agent field, citing them as “common algorithms” that any team would adopt.

Some observers noted that EvoMapAI has previously been involved in similar disputes, suggesting the accusations might also serve a publicity strategy.

Following the controversy, EvoMapAI obfuscated its core modules and switched its license from MIT to GPL‑3.0, issuing a statement that disrespect for open‑source code is becoming a widespread problem across developers, startups, and large enterprises.

The author concludes that while similarity is unsurprising in a rapidly converging field, the detailed term‑by‑term matches presented by EvoMapAI are hard to dismiss as mere parallel evolution, and the founder’s “Delete your account” reply damages the project’s reputation. The episode highlights the growing ambiguity of attribution boundaries in the booming AI tooling ecosystem and underscores the importance of explicit credit when building on open‑source work.

Original Source

Signed-in readers can open the original source through BestHub's protected redirect.

Sign in to view source
Republication Notice

This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactadmin@besthub.devand we will review it promptly.

software architectureAI agentsopen-sourcecode plagiarismself-evolution
AI Insight Log
Written by

AI Insight Log

Focused on sharing: AI programming | Agents | Tools

0 followers
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.