#1: cathrynlavery/diagram-design
HTML · 19,954 stars total, +15,600 this period
The project cathrynlavery/diagram-design provides a collection of 29 self-contained diagram types for use with Claude Code. These diagrams are implemented purely in HTML and SVG, ensuring simplicity and minimalism without the inclusion of shadow effects or overly complex configurations often associated with similar tools like Mermaid. The sheer number of stars accumulated—15,600 within this period—and a total star count of 19,954 indicate a significant community interest and utility in this project. Given its straightforward implementation and broad applicability for creating clear, uncluttered diagrams directly through HTML, it's likely trending due to the increasing demand for clean, customizable diagram solutions without additional dependencies or complex setup processes.
#2: semantica-agi/semantica
Python · 8,327 stars total, +5,284 this period
The project semantica-agi/semantica is a graph-native infrastructure designed to support context and accountable AI systems. It addresses the need for robust, transparent, and interconnected AI solutions that can operate effectively within complex data environments. Given its recent surge in popularity, with an impressive gain of 5,284 stars over a relatively short period, it's clear that many developers are finding value in this tool. The rapid increase in star count suggests that the project has effectively met a significant demand for more open and transparent AI systems, likely due to its unique approach to leveraging graph technology.
#3: PrimeIntellect-ai/prime-agent
TypeScript · 16,701 stars total, +6,435 this period
Prime Agent is an open-source coding and research agent built for long-running, autonomous work. The "RLM" in its tagline stands for Recursive Language Model: instead of stuffing everything into one prompt, the agent treats context as variables and calls tools and sub-agents like functions inside a persistent REPL. Sub-agents are built in, so it can spawn child agents for parallel or background work, and it can re-enter a session on a schedule or heartbeat.
It picked up 6,435 stars this period for a total of 16,701, which fits the wider interest in agent harnesses that can keep working on a task for hours rather than minutes.
#4: NVIDIA-NeMo/Switchyard
Rust · 1,730 stars total, +1,435 this period
The project NVIDIA-NeMo/Switchyard is a Rust-based tool designed to facilitate the management of Large Language Model (LLM) applications. It allows users to route traffic across different models and providers while maintaining compatibility with native APIs from OpenAI and Anthropic, thus providing a flexible solution for selecting, benchmarking, and optimizing costs and performance in LLM deployments.
Given its recent surge in popularity, as indicated by the significant increase of 1,435 stars over the latest period, Switchyard is likely trending due to its unique capability to simplify model management across multiple providers. The high number of stars (now at 1,730) suggests that it addresses a pressing need for developers looking to streamline their LLM applications and optimize resource usage without compromising on API compatibility.
#5: megadose/holehe
Python · 13,336 stars total, +1,287 this period
holehe is a Python OSINT tool that checks whether an email address is registered on sites such as Twitter and Instagram. It works by using each site's forgotten-password and sign-up flows to see whether the address is recognised, and it reports whatever extra information those flows give back. It is an older, well-established project rather than a new release: 13,336 stars in total, with 1,287 of them gained this period, so this week's appearance is a resurgence. The usual use is security auditing and checking your own exposure.