Awesome Self-Evolving AI Agents for Hermes Agent

Awesome Self-Evolving AI Agents

Shiyao-Huang/awesome-agent-evolution

Open survey and evidence map for self-evolving AI agents, memory, skills, harnesses and benchmarks

In short

Awesome Self-Evolving AI Agents is an open survey and evidence map on how AI agents improve themselves, covering memory, skills, harnesses, benchmarks and agent swarms. It tracks Hermes Agent among the projects it examines, as an example of a usable product-style agent.

What Awesome Self-Evolving AI Agents does

The repository is written mainly in Chinese, with an English version in README-EN.md and a companion site at agent-evolution.com. Its central test asks five questions of any system that claims to evolve: what changed, why, who verified it, whether the change was kept, and whether it can be rolled back. It frames this as an Observe, Interpret, Modify, Verify, Retain loop and sorts systems into five kinds of evolution loop: spec to execution, search, evaluator, reflection and memory, and population or archive.

Supporting material includes a code self-improvement benchmark matrix, per-project reports and model cards, a paper draft, and an Evolve-AGI Index. The README calls the index a working prototype with editorial, proposed weights, and says it is not an AGI score. It also publishes periodic metadata updates for tracked repositories, and NousResearch/hermes-agent appears in those updates.

Key features

Definition of a self-evolving agent built on five checks and a five-step feedback loop
Five evolution-loop categories for comparing systems
Code self-improvement benchmark matrix
Project reports with evidence status for each tracked repository
Evolve-AGI Index prototype for weighing benchmark, loop and governance evidence
English entry pages under /en/ on the companion site

When to use it

Judge whether an agent project really improves from feedback or only looks good in a demo
Find research on memory, skill and harness evolution to compare with Hermes Agent's learning loop
Pick agent projects to study using evidence rather than popularity

Who it is for: Researchers and engineers who want a structured way to evaluate self-improving agent systems, Hermes Agent included.

How it fits with Hermes Agent

It is a general research survey and not built for Hermes; Hermes Agent appears as one of the tracked projects.

Note: Many long-tail pages are still Chinese-first, and the README says it does not claim full English translation parity.

FAQ

What is Awesome Self-Evolving AI Agents?

It is an open survey and evidence map on AI agent self-evolution. It helps readers decide whether a system genuinely improves from feedback, and it organizes projects, papers, benchmarks and reports.

Does Awesome Self-Evolving AI Agents work with Hermes Agent?

It is not software that runs alongside Hermes. It is a survey that includes NousResearch/hermes-agent among the projects it tracks.

Is Awesome Self-Evolving AI Agents free and open source?

Yes. The repository is published under the MIT license.