On April 15, 2026, a major controversy erupted in the open‑source AI agent field. Silicon Valley lab Nous Research's flagship project Hermes Agent (over 89k stars on GitHub) was publicly accused by Chinese AI team EvoMap of being highly isomorphic to EvoMap's Evolver engine, which was open‑sourced 36 days earlier. EvoMap alleges the core "self‑improving" functionality shows one‑to‑one correspondence with a 10‑step main loop, systematic substitution across 12 sets of core terms, and no mention of the original authors across seven public materials. EvoMap published a detailed technical comparison report that shocked the open‑source community and was dubbed "another scandal in the open‑source world."
Hermes Agent, developed by Nous Research and billed as "the agent that grows with you," centers on a built‑in learning loop that can automatically distill skills from usage, achieve cross‑session memory, and perform self‑optimization. Since its release in late February 2026, the project quickly gained attention for its lightweight design, persistent memory, and multi‑model support, becoming one of the most watched open‑source agent frameworks after OpenClaw. Many developers hailed it as a representative of the "next‑generation intelligent agent," praising its capability to become "smarter with use."
However, EvoMap's accusations target Hermes Agent's core competitive feature — the self‑improving mechanism. In its report EvoMap lists multiple pieces of evidence:
- Highly consistent 10‑step main loop: From task execution and result evaluation to skill extraction, memory updates and self‑optimization, the processes are nearly identical.
- Systematic term substitution: Key concepts in Evolver (such as "experience accumulation," "skill generation," and "hierarchical memory") were replaced with near‑synonyms in Hermes Agent, while the underlying logic and data structures closely match.
- Zero attribution: Hermes Agent's README, documentation, blog posts, and code comments make no reference or acknowledgment of Evolver or the EvoMap team.
EvoMap emphasized that the allegation is not merely "code copying," but architectural isomorphism — a nearly identical self‑improving closed loop apparently reproduced without access to Evolver's source. Such behavior is seen in the open‑source community as a serious breach of the basic ethics of respecting prior art and citing sources.
Nous Research's response intensified the controversy
Faced with the detailed comparison, responses from the Hermes Agent team and Nous Research founder Teknium fueled further discontent. Teknium replied on X: "Delete your account.", and called the accusations "completely brainless." The Hermes team later removed some related discussion posts but offered no technical rebuttal or revision history.
The reply was widely criticized as "arrogant" and evasive. The open‑source community generally holds that even in cases of clean‑room reimplementation, teams should at minimum acknowledge inspiration and thank predecessors. Some developers argued that Hermes Agent's rapid rise neglected respect for earlier open‑source work, contradicting the spirit of open source.
Community shock and reflection
The incident quickly spread across GitHub, Reddit, Zhihu and X. Many developers said Hermes Agent's self‑improving feature had been its key selling point; the plagiarism allegations damaged the project's reputation. Some users paused using Hermes and migrated to other frameworks; others called on Nous Research to publish change logs or add acknowledgments.
The case also prompted broader discussion about open‑source ethics:
- Voices of Chinese open‑source projects: As a smaller team with limited resources, EvoMap open‑sourced its work first but alleges a Silicon Valley high‑profile project adopted its ideas without attribution, highlighting unequal visibility and voice in the open‑source community.
- Questioning the spirit of open source: Open source is not just about releasing code; it also requires respecting predecessors, citing sources and maintaining community trust. The Hermes Agent incident is seen as a "crisis of trust" that could affect future collaboration willingness.
- Impact on the AI agent track: Hermes Agent had been seen as the most promising framework after OpenClaw; the controversy may accelerate developers' demand for transparency and clear provenance. Future open‑source releases in this space may place greater emphasis on citations and acknowledgments.
Industry lesson: open source needs stricter ethical norms
The Hermes Agent case is not unique. Past incidents such as the Claude Code leak have already sparked disputes over clean‑room rewrites. This episode is another reminder that in the fast‑iterating world of AI agents, technological innovation must rest on respect for prior art. Otherwise, high star counts and popularity cannot mask fractures of trust.
For Nous Research, how it responds — whether by adding acknowledgments or amending documentation — will directly affect the project's long‑term reputation. For smaller teams like EvoMap, this is also a crucial moment to demonstrate technical capability and attract community attention.
Another open‑source scandal underlines that technology can be copied quickly, but respect and integrity cannot be rushed. At a key stage when AI agents are evolving from "tools" into "intelligent partners," a transparent, compliant and trust‑based open‑source culture matters more than mere model performance.
The Hermes Agent incident may be only the beginning. As more high‑performance agent projects emerge, balancing innovation speed with ethical boundaries will be a long‑term challenge the open‑source community must collectively address.
▶️ Enjoy this piece? Creating content is hard!
Follow our YouTube channel:
AI Rosetta
Your likes, subscriptions and shares are our greatest support! 🙏
More AI news at: https://www.airosetta.com