Awesome-Self-Evolving-Agents: Explore Self-Evolving Agent Research

Overview

Awesome-Self-Evolving-Agents is a curated research collection built around the survey “A Systematic Survey of Self-Evolving Agents: From Model-Centric to Environment-Driven Co-Evolution.” It brings together work on agents that improve their models, knowledge, experience, tools, architectures, or environments, helping readers navigate a broad and fast-moving field.

The repository is aimed at researchers, students, and engineers who want to survey the area or find relevant papers, evaluation benchmarks, libraries, and application examples. Its taxonomy groups material by how agents evolve, rather than presenting a flat reading list.

Key Features

• Organizes research into model-centric, environment-centric, and model-environment co-evolution.
• Subdivides topics such as self-correction, online learning, memory, skill evolution, and agent topology.
• Links to related surveys and a broad collection of research papers.
• Catalogs benchmarks for self-evolution, general capabilities, and agentic reasoning.
• Lists open-source libraries for orchestration, training, alignment, fine-tuning, and serving.
• Collects application examples in areas including scientific discovery and software engineering.
• Invites contributions through GitHub issues and pull requests.

Use Cases

• Researchers entering the field can use the taxonomy and related surveys to identify key subtopics and starting papers.
• Graduate students preparing a literature review can follow the topic structure to compare approaches across model, environment, and co-evolution.
• Agent developers choosing evaluations can scan the benchmark listings for options covering reasoning, web navigation, coding, or self-improvement.
• Teams exploring agent infrastructure can use the library directory to locate projects for orchestration, training, or inference.

Getting Started

Open the repository README and use its table of contents to browse the taxonomy, papers, benchmarks, libraries, and applications. There is no installation step described.

License in Plain Words

No standard license was detected. Without a license, the default copyright rules apply: you may read the code, but not reuse it. Check the repository or ask the authors.

Alternatives

• Awesome-Dynamic-Agent-Skills: Focuses on dynamic, self-evolving skill systems and their lifecycle, while this project surveys self-evolving agents more broadly.

Considerations

The collection is actively updated, but it has no published releases, so entries and organization may change over time. The activity data shows recent commits and merged pull requests, while no issues were closed in the measured recent period; readers should not assume issue-response times. The README invites contributions, but inclusion in a curated list is not an independent assessment of a paper, benchmark, or project. No license is specified in the supplied repository metadata, so check the repository before reusing its contents.