GIT_FEED
WEEKLYWeek of Apr 5, 2026

// RISING THIS WEEK

The 10 fastest-accelerating repos on GitHub for the week of Apr 5, 2026. Ranked by 7-day star velocity, filtered to projects with enriched context.

121.6k
Stars gained
10
Repos ranked
+5%
Avg WoW accel
1
msitarzewski/agency-agents

Agency Agents is a library of pre-built AI assistant personalities — each designed for a specific job like frontend development, backend architecture, or community management — that you can plug into AI coding tools to get expert-level help on demand. Instead of writing your own instructions for AI tools from scratch, you pick a specialist from the roster and instantly have a capable, focused AI collaborator.

42Hot

// why it matters As AI tools become standard in every product team's workflow, the teams that get the most value will be those with the best 'playbooks' for directing AI — and this project hands you a ready-made roster of those playbooks for free. With 30,000 stars, it signals strong market demand for structured, role-based AI workflows, which is a pattern worth understanding whether you're building products with AI or building AI products.

+22.9k stars this week+0% WoW74.5k total stars11.7k forksShell
2
666ghj/MiroFish

MiroFish is an open-source engine that uses large groups of AI agents working together — like a digital swarm — to make predictions about almost anything, from financial markets to public opinion and social trends. Think of it as a crowd-wisdom system where many AI 'minds' collaborate and share knowledge to forecast outcomes more accurately than a single model could.

38Active

// why it matters With over 30,000 stars, this project signals massive developer appetite for prediction infrastructure that goes beyond single AI models — suggesting a market shift toward multi-agent forecasting as a core product capability. Founders and investors building in fintech, market research, or social analytics should take note: accessible swarm-based prediction tools could soon become a competitive baseline rather than a differentiator.

+20.2k stars this week+0% WoW51.3k total stars7.6k forksPython
3
koala73/worldmonitor

World Monitor is a free, open-source intelligence dashboard that pulls together news from hundreds of sources, live maps, and financial signals into a single screen, giving users a real-time picture of global events and risks. It uses AI to summarize and connect the dots across geopolitical, economic, and infrastructure developments, and can run entirely on your own computer without sending data to the cloud.

41Hot

// why it matters With nearly 35,000 stars, this project signals massive demand for affordable, self-hosted alternatives to expensive enterprise intelligence platforms like Palantir — a clear market gap that founders building in the security, media, or risk-intelligence space should pay attention to. For product teams, it demonstrates that users will flock to open-source tools that bundle AI summarization, geospatial context, and real-time data in one place, especially when the incumbent solutions cost a fortune.

+13.6k stars this week+0% WoW47.2k total stars7.6k forksTypeScript
4
ruvnet/RuView

RuView uses ordinary WiFi signals to detect human presence, movement, and even vital signs like heart rate and breathing — all without cameras, wearables, or an internet connection. It runs on cheap hardware (around $1 per sensor node) and learns the layout of a space on its own over time, getting smarter the longer it operates.

38Active

// why it matters This opens up a massive market for privacy-first sensing in smart buildings, elder care, retail analytics, and security — anywhere cameras are too intrusive or expensive to deploy at scale. Builders can now add human-presence awareness to physical spaces using off-the-shelf hardware, without storing a single image or relying on cloud infrastructure.

+11.7k stars this week+0% WoW46.0k total stars6.2k forksRust
5
moeru-ai/airi

Project AIRI is an open-source platform for creating self-hosted AI companions — think interactive virtual characters powered by large language models that can hold real-time voice conversations, play games like Minecraft and Factorio, and be animated with lifelike 3D avatars. It's essentially a toolkit for building your own version of Neuro-sama, the popular AI-powered virtual streamer, running entirely on infrastructure you own and control.

38Active

// why it matters With over 31,000 stars, this project signals massive consumer appetite for interactive AI companions that go beyond simple chatbots — opening product opportunities in entertainment, gaming, and digital companionship without depending on closed platforms like Character.ai. For founders and investors, it represents a foundational layer for the emerging 'digital beings' market, where owning the character and the underlying system is a meaningful competitive differentiator.

+10.7k stars this week+0% WoW37.4k total stars3.7k forksTypeScript
6
Yeachan-Heo/oh-my-claudecode

Oh My Claude Code lets multiple AI coding assistants work simultaneously on different parts of a software project, coordinating their efforts like a team rather than a single assistant working alone. It's built on top of Anthropic's Claude Code tool and essentially turns one AI helper into an organized crew of AI workers tackling tasks in parallel.

42Hot

// why it matters As AI coding tools move from novelty to core development infrastructure, the ability to run coordinated teams of AI agents could dramatically compress software build times — a meaningful competitive advantage for any startup or product team. With nearly 15,000 stars on GitHub, this level of community interest signals that multi-agent AI development workflows are quickly becoming a mainstream expectation, not an edge case.

+9.9k stars this week+46% WoW25.7k total stars2.4k forksTypeScript
7
obra/superpowers

Superpowers is a plug-in framework for AI coding assistants that makes them work more like a disciplined software team — slowing down to clarify requirements, writing a clear plan, and then executing step by step with minimal supervision. Instead of an AI that jumps straight to writing code and goes off the rails, you get one that follows a structured process from idea to finished, tested software.

37Active

// why it matters As AI coding tools become a core part of how software gets built, the difference between an AI that produces unreliable output and one that can work autonomously for hours on a real project is enormous for team productivity and product velocity. With 130,000 stars, this project signals strong market demand for structured AI development workflows — a space where tooling, best practices, and even new business models are still being established.

+8.8k stars this week+0% WoW139.4k total stars11.8k forksShell
8
D4Vinci/Scrapling

Scrapling is a Python tool that automatically collects data from websites at scale, and it's smart enough to keep working even when those websites change their layout or try to block automated visitors. Think of it as a self-healing data collection robot that can quietly gather information from across the web without getting shut out.

39Active

// why it matters For any product that depends on external web data — pricing intelligence, market research, lead generation, or competitive monitoring — this dramatically reduces the engineering effort and ongoing maintenance cost of keeping those data pipelines alive. With over 22,000 stars on GitHub, it signals strong market demand for resilient, low-friction web data collection, which is increasingly a competitive advantage across industries.

+8.1k stars this week+0% WoW35.0k total stars2.9k forksPython
9
anthropics/skills

This is Anthropic's official library of pre-built capabilities that can be plugged into Claude, their AI assistant, allowing it to perform specific tasks like browsing the web, writing code, or analyzing files. Think of it as an app store of abilities for AI agents — reusable building blocks that developers can mix and match to create AI-powered workflows without starting from scratch.

27Active

// why it matters With over 107,000 stars, this is one of the most-watched repositories on GitHub, signaling that AI agents that can take real-world actions are becoming a core product category rather than a research curiosity. Builders who understand this ecosystem early can design products around AI that doesn't just answer questions, but actually does things — a fundamental shift in what software can offer customers.

+7.9k stars this week+0% WoW112.4k total stars12.7k forksPython
10
shanraisshan/claude-code-best-practice

This project is a community-curated guide for getting the most out of Claude Code, Anthropic's AI-powered coding assistant, covering advanced features like automated checkpointing, plugin systems, and letting the AI control your computer screen. Think of it as an unofficial handbook that helps teams use this AI coding tool far more effectively than the official documentation alone would allow.

37Active

// why it matters With over 30,000 stars, this repository signals massive market demand for AI coding assistants and the meta-skill of knowing how to direct them effectively — a capability gap that's becoming a real competitive advantage for software teams. Founders and PMs should note that 'how to work with AI agents' is fast becoming its own discipline, and early teams that master it will ship products significantly faster than those who don't.

+7.8k stars this week+0% WoW32.8k total stars3.0k forksHTML

Methodology

Repos are ranked by 7-day star velocity — the number of new stars gained in the last 7 days. Only repos with enriched context (Claude-generated summaries and Early Signal Scores) are included. Week-over-week acceleration compares this week's star gain to the previous week's. Data is sourced from daily GitHub API snapshots and refreshed every hour.

// SUBSCRIBE

The repos that moved this week, why they matter, and what to watch next. One email. No noise.