elder-plinius/CL4R1T4S
CL4R1T4S is a project that provides transparency by collecting and sharing the hidden system prompts and guidelines used by major AI models and agents to promote trust and understanding of AI behavior.
Awesome AI Agents › LLM Research Repositories
WecoAI/awesome-autoresearch
Awesome AutoResearch is a curated index of projects built on AutoResearch, the automated optimization loop released by Andrej Karpathy as a single markdown file called program.md. That file instructs a coding agent such as Claude Code or Codex to repeatedly edit one target file, run it for a fixed time budget on a GPU, check whether an evaluation metric improved, and then either commit the change or revert it, looping indefinitely. The original program.md targets training a GPT model, but the list makes the case that the structure generalizes: iteratively optimizing a file against a metric with a keep-or-discard loop is portable to other domains, and only the wording of program.md changes. The distinguishing rule of this list is that every entry links to the optimization trajectory itself, not only the final result, so a reader can see which changes the agent tried and which it rejected. Entries are presented as a table with the use case, a short description of the outcome, the author, links to the code or write-up, and the trace. Documented adaptations include the original nanoGPT training optimization, a run against the Liquid template engine that produced dozens of automated commits with measured speed and allocation improvements, CUDA kernel optimization, voice-agent prompt tuning, sports prediction models built from biomechanics data, tabular machine learning with XGBoost, and reinforcement learning hyperparameter search. At least one entry is included specifically because the author encountered and documented reward hacking, where the agent improved the metric without improving the underlying task. The repository is maintained by WecoAI, carries the standard Awesome badge, and has no code of its own. It is useful to anyone evaluating self-improving agent loops who wants concrete evidence of what such runs actually produce and where they fail.
https://github.com/WecoAI/awesome-autoresearch
Also in LLM Research Repositories
WooooDyy/LLM-Agent-Paper-List
A comprehensive repository compiling essential academic papers and a detailed survey on the rise, framework, applications, and societal aspects of Large Language Model based AI agents.
mahseema/awesome-ai-tools
A curated repository listing top Artificial Intelligence tools across various categories including generative AI, chatbots, search engines, and writing assistants.
filipecalegario/awesome-generative-ai
A curated and comprehensive list of Generative AI tools, projects, models, artworks, and educational resources, regularly updated to reflect the latest advancements in the field.
DSXiangLi/DecryptPrompt
DecryptPrompt is a comprehensive resource hub that summarizes prompt and LLM research papers, open-source datasets and models, and AIGC applications, providing in-depth tutorials and analyses for AI researchers and practitioners.
zjunlp/LLMAgentPapers
LLMAgentPapers is a curated repository of essential academic papers on Large Language Model agents, covering topics from agent personality to multi-agent systems and applications.
luo-junyu/Awesome-Agent-Papers
A comprehensive and up-to-date research collection of papers on Large Language Model (LLM) agents, covering methodologies, applications, challenges, and key categories such as construction, collaboration, evolution, tools, security, benchmarks, and real-world applications.
webfuse-com/awesome-autoresearch
Curated index of autonomous improvement loops and research-agent systems descended from karpathy/autoresearch, grouped into general-purpose forks, research systems, platform ports, domain adaptations and benchmarks.