Today's AI Research: Fastest-Growing Projects — April 28, 2026
Today's AI research, we've seen a surge of interest in tools that facilitate the development and evaluation of large language models (LLMs) and autonomous AI agents. Researchers are also exploring new applications for LLMs, such as trading analysis and medical research. Meanwhile, security concerns around API relay services and exposed LLM endpoints have come to the forefront.
Openedclaude's claude-reviews-claude repository has taken the top spot with a growth score of 64.45 and 1,362 stars. This project provides an in-depth architectural analysis of Claude Code v2.1.88, featuring a 17-chapter bilingual deep dive into its source code. Its rapid growth can be attributed to the increasing popularity of LLMs and the need for researchers to understand their underlying mechanics.
AutoMedBench's AutoMedBench repository has seen significant activity with 42 commits in the past 30 days, earning it a growth score of 23.08 and 22 stars. This medical auto-research benchmark is designed for autonomous AI agents, providing a standardized framework for evaluating their performance in various medical tasks. Its growing popularity reflects the expanding scope of AI research in healthcare.
Toby-bridges' api-relay-audit repository has gained traction with 70 commits in the past month, resulting in a growth score of 13.03 and 255 stars. This security audit tool detects hidden prompt injection, prompt leakage, instruction override, and context truncation in third-party AI API relay services. As more researchers rely on these services, the need for robust security measures has driven interest in this project.
Matrix-agent's awesome-agentic-world-modeling repository boasts a growth score of 10.88 and 69 stars, despite only three commits in the past month. This curated list provides foundations, capabilities, laws, and beyond for agentic world modeling, offering valuable resources for researchers exploring this area. Its popularity stems from the growing recognition of agentic world modeling as a crucial aspect of AI research.
Gameworld-project's gameworld repository has seen 11 commits in the past month, earning it a growth score of 9.23 and 156 stars. This project aims to establish standardized and verifiable evaluation methods for multimodal game agents, addressing a pressing need in the field. Its growing interest reflects the expanding scope of AI research in gaming and interactive environments.
7WaySecurity's ai_osint repository has gained attention with 40 commits in the past month, resulting in a growth score of 8.80 and 70 stars. This curated list provides resources for discovering exposed LLM endpoints, leaked AI API keys, misconfigured vector databases, and unprotected AI agents using Google dorks, Shodan queries, GitHub dorks, and other techniques. As security concerns around AI grow, this repository has become an essential resource.
Thunlp's OPD repository boasts a growth score of 7.97 and 156 stars, with 14 commits in the past month. This official repository for the paper "Rethinking On-Policy Distillation of Large Language Models" offers insights into the phenomenology, mechanism, and recipe for on-policy distillation. Its popularity reflects the ongoing interest in optimizing LLM performance.
AMAP-ML's DCW repository has seen seven commits in the past month, earning it a growth score of 7.18 and 110 stars. This project elucidates the SNR-t bias of diffusion probabilistic models, addressing a key challenge in AI research. Its growing interest stems from the increasing recognition of the importance of understanding these biases.
Zubair-trabzada's ai-trading-claude repository has gained traction with two commits in the past month, resulting in a growth score of 6.40 and 83 stars. This AI trading research engine for Claude Code analyzes stocks, options strategies, sector rotation, portfolio analysis, and PDF reports using 16 skills and five parallel agents. Its popularity reflects the growing interest in applying LLMs to financial analysis.
Gloriaameng's Awesome-Agent-Harness repository boasts a growth score of 5.98 and 89 stars, with 35 commits in the past month. This survey provides a taxonomy of agent harness engineering for large language model agents, analyzing over 110 papers and 23 systems. Its popularity stems from the growing recognition of the importance of agent harnessing in AI research.
Note: The remaining tools were skipped due to lack of meaningful descriptions or insufficient data.
Openedclaude's claude-reviews-claude repository has taken the top spot with a growth score of 64.45 and 1,362 stars. This project provides an in-depth architectural analysis of Claude Code v2.1.88, featuring a 17-chapter bilingual deep dive into its source code. Its rapid growth can be attributed to the increasing popularity of LLMs and the need for researchers to understand their underlying mechanics.
AutoMedBench's AutoMedBench repository has seen significant activity with 42 commits in the past 30 days, earning it a growth score of 23.08 and 22 stars. This medical auto-research benchmark is designed for autonomous AI agents, providing a standardized framework for evaluating their performance in various medical tasks. Its growing popularity reflects the expanding scope of AI research in healthcare.
Toby-bridges' api-relay-audit repository has gained traction with 70 commits in the past month, resulting in a growth score of 13.03 and 255 stars. This security audit tool detects hidden prompt injection, prompt leakage, instruction override, and context truncation in third-party AI API relay services. As more researchers rely on these services, the need for robust security measures has driven interest in this project.
Matrix-agent's awesome-agentic-world-modeling repository boasts a growth score of 10.88 and 69 stars, despite only three commits in the past month. This curated list provides foundations, capabilities, laws, and beyond for agentic world modeling, offering valuable resources for researchers exploring this area. Its popularity stems from the growing recognition of agentic world modeling as a crucial aspect of AI research.
Gameworld-project's gameworld repository has seen 11 commits in the past month, earning it a growth score of 9.23 and 156 stars. This project aims to establish standardized and verifiable evaluation methods for multimodal game agents, addressing a pressing need in the field. Its growing interest reflects the expanding scope of AI research in gaming and interactive environments.
7WaySecurity's ai_osint repository has gained attention with 40 commits in the past month, resulting in a growth score of 8.80 and 70 stars. This curated list provides resources for discovering exposed LLM endpoints, leaked AI API keys, misconfigured vector databases, and unprotected AI agents using Google dorks, Shodan queries, GitHub dorks, and other techniques. As security concerns around AI grow, this repository has become an essential resource.
Thunlp's OPD repository boasts a growth score of 7.97 and 156 stars, with 14 commits in the past month. This official repository for the paper "Rethinking On-Policy Distillation of Large Language Models" offers insights into the phenomenology, mechanism, and recipe for on-policy distillation. Its popularity reflects the ongoing interest in optimizing LLM performance.
AMAP-ML's DCW repository has seen seven commits in the past month, earning it a growth score of 7.18 and 110 stars. This project elucidates the SNR-t bias of diffusion probabilistic models, addressing a key challenge in AI research. Its growing interest stems from the increasing recognition of the importance of understanding these biases.
Zubair-trabzada's ai-trading-claude repository has gained traction with two commits in the past month, resulting in a growth score of 6.40 and 83 stars. This AI trading research engine for Claude Code analyzes stocks, options strategies, sector rotation, portfolio analysis, and PDF reports using 16 skills and five parallel agents. Its popularity reflects the growing interest in applying LLMs to financial analysis.
Gloriaameng's Awesome-Agent-Harness repository boasts a growth score of 5.98 and 89 stars, with 35 commits in the past month. This survey provides a taxonomy of agent harness engineering for large language model agents, analyzing over 110 papers and 23 systems. Its popularity stems from the growing recognition of the importance of agent harnessing in AI research.
Note: The remaining tools were skipped due to lack of meaningful descriptions or insufficient data.