Preset
Background
Text
Font
Size
Width
Account Tuesday, March 17, 2026

The Git Times

“We live in a society exquisitely dependent on science and technology, in which hardly anyone knows anything about science and technology.” — Carl Sagan

AI Models
Claude Sonnet $15/M GPT-4.1 $8/M Gemini 2.5 Pro $10/M Grok 3 $0.50/M DeepSeek V3 $0.75/M Llama 4 Maverick $0.60/M
Full Markets →

Attention Residuals Let Transformers Selectively Tap Deeper Layer Wisdom

MoonshotAI's AttnRes swaps uniform residuals for smart attention over prior outputs, fixing dilution in massive models

In the relentless quest to build ever-deeper Transformers, a breakthrough has emerged from MoonshotAI: Attention Residuals (AttnRes), a simple yet profound drop-in replacement for standard residual connections. This innovation allows each Transformer layer to dynamically attend to all preceding layers' representations, using learned, input-dependent attention weights. No more blindly summing everything with fixed unit weights—AttnRes empowers layers to selectively aggregate what's relevant from depth.

The core problem it tackles is painfully familiar to model builders. Traditional residuals in PreNorm Transformers—h_l = norm(h_{l-1} + layer(h_{l-1}))—accumulate outputs uniformly. As layers stack up (think 100+ in modern LLMs), early layers' contributions get diluted, while hidden-state magnitudes explode unboundedly. This leads to training instability, vanishing gradients for shallow features, and suboptimal performance at scale. AttnRes fixes this elegantly:

$ \mathbf{h}l = \sum{i=0}^{l-1} \alpha_{i \to l} \cdot \mathbf{v}_i $

Here, softmax-normalized weights $\alpha_{i \to l}$ come from dot-product attention between a single learned pseudo-query $\mathbf{w}_l \in \mathbb{R}^d$ per layer and keys from prior outputs. It's content-aware: a deep layer might heavily weight a mid-layer's edge detector if the input demands it. Computationally, it's cheap—a linear projection and softmax per layer—with Full AttnRes storing all prior outputs for O(Ld) memory, where L is layers and d is dimension.

For production scale, Block AttnRes shines brighter. It groups layers into N blocks (~8 suffices), using standard residuals within blocks and attention only over block summaries. Memory drops to O(Nd), with "marginal overhead" and near-Full AttnRes gains. PyTorch pseudocode illustrates the plug-and-play vibe:

def block_attn_res(blocks: list[Tensor], partial_block: Tensor, proj: Linear, norm: RMSNorm) -> Tensor:
    # Inter-block attention over block reps + partial sum
    # ... (full impl in repo)

Results? The repo's benchmarks show AttnRes boosting accuracy on language modeling and vision tasks, especially beyond 50 layers, where baselines falter. It's a universal upgrade for any Transformer stack—GPT-style LLMs, ViTs, even diffusion models—without retraining from scratch.

What makes AttnRes technically mesmerizing is its minimalism: one parameter vector per layer unlocks "depth awareness." No new architectures, no massive compute—just smarter residuals. Developers are buzzing because it democratizes ultra-deep models: swap in AttnResBlock and watch perplexity plummet.

Already gaining explosive traction in just days, MoonshotAI/Attention-Residuals is catnip for the Transformer tinkerer. It ties into arXiv paper details, full evals, and citations, positioning it as the next must-try for pushing model frontiers. For builders weary of depth's diminishing returns, AttnRes isn't hype—it's the residual evolution Transformers have craved.

(Word count: 462)

Use Cases
  • LLM engineers training 100-layer models without gradient dilution.
  • Vision Transformer devs stabilizing deep ViTs for better accuracy.
  • Diffusion model builders enhancing long-sequence generation stability.
Similar Projects
  • bzhangGo/rmsnorm - Stabilizes norms in deep nets but ignores residual dilution.
  • microsoft/DeepNorm - Improves training dynamics via scaling, lacks selective depth attention.
  • lucidrains/deepnorm-pytorch - Focuses on post-norm tweaks, not input-aware aggregation.

More Stories

MoonBit Tool Runs GitHub Actions Workflows Locally

actrun offers CLI-compatible execution, action skipping and Nix integration for faster debugging

mizchi/actrun · MoonBit · 458 stars

Claude Plugin Builds Interactive Knowledge Graphs from Codebases

Lum1104/Understand-Anything uses multi-agent analysis for visual exploration of files, functions and dependencies

Lum1104/Understand-Anything · TypeScript · 575 stars

NVIDIA NemoClaw Simplifies Secure OpenClaw Agent Deployment

TypeScript plugin installs OpenShell runtime for sandboxed autonomous agents on Ubuntu

NVIDIA/NemoClaw · TypeScript · 1.9k stars

GitHub Repo Exposes Meta's Lobbying for App Store Burdens

Open-source OSINT tracks influence operation using public records to evade social media regulations

Open-Source Skills Ecosystems Supercharge Modular AI Agents

Devs build plug-and-play tools for browser hacks, trading bots, and email mastery, fueling composable agent architectures

trend/ai-agents · Trend · 0 stars

Open Source Booms with LLM Agent Skills, Proxies, and Toolchains

Modular skills libraries, CLI proxies, and inference optimizers signal a shift to composable, hardware-agnostic agent ecosystems.

trend/llm-tools · Trend · 0 stars

Agentic Web Frameworks Rise: AI Agents Reshape Open Source Development

Modular tools blend natural language control, performant backends, and UI components for autonomous web app building and management.

trend/web-frameworks · Trend · 0 stars

Deep Cuts

Use Cases
  • Indie hackers automating client email follow-ups and replies.
  • Dev teams building AI triage for support inboxes daily.
  • Builders creating Claude agents for personalized outreach campaigns.
Similar Projects
  • LangChain - Broader agent orchestration, lacks email-specific MCP depth.
  • Auto-GPT - Autonomous tasks, no native Claude email integration.
  • CrewAI - Multi-agent collab, less focus on comms tools like inbox/SMS.
Use Cases
  • UI designers extracting competitor logos and colors for mood boards.
  • Frontend devs pulling site palettes to theme client dashboards quickly.
  • Marketers automating brand asset collection for campaign mockups.
Similar Projects
  • Brand.dev - proprietary SaaS; OpenBrand is free, self-hosted OSS
  • color-thief - colors only; lacks logos, backdrops extraction
  • real-favicon-generator - favicon-centric; ignores full brand assets

Quick Hits

OpenBrand Extracts logos, colors, and backdrops from any website, giving builders a free tool to snag brand assets instantly. 474
jobs Visually explores BLS job data interactively, letting builders dive into occupational trends without spreadsheets. 565
ragflow Fuses RAG with agent smarts for superior LLM context layers, supercharging open-source AI apps. 75.2k
MANPADS-System-Launcher-and-Rocket Designs a MANPADS launcher and rocket system, enabling builders to model portable air defense hardware. 1.4k
OpenSquirrel Runs Claude Code, Codex, Cursor, and OpenCode side-by-side in Rust for distraction-free AI coding multitasking. 873
llm-architecture-gallery Supplies source data for LLM architecture gallery, helping builders dissect and visualize transformer designs. 612
OpenGenerativeUI Builds generative UIs with AI-driven components, empowering devs to create dynamic interfaces effortlessly. 559
prompt-master Crafts perfect prompts for any AI tool via Claude skill, slashing token waste and re-prompts for pros. 425

Microsoft's 12-Lesson Course Unlocks AI Agent Building for Developers

Jupyter-based tutorials demystify agentic frameworks like AutoGen and Semantic Kernel for hands-on agent creation.

microsoft/ai-agents-for-beginners Jupyter Notebook 54.2k stars

Developers diving into AI agents face a steep curve: orchestrating LLMs for autonomous tasks requires mastering frameworks, reasoning loops, and retrieval-augmented generation. Microsoft's ai-agents-for-beginners repo cuts through this with 12 self-contained Jupyter Notebook lessons, delivering practical blueprints for agentic systems.

The course starts with agent fundamentals—defining roles, tools, and multi-agent collaboration—then scales to advanced patterns. Lessons unpack AutoGen, Microsoft's open-source framework for conversational agents, showing how to spin up dynamic groups where agents delegate tasks, critique outputs, and iterate via human-in-loop proxies. Code snippets demonstrate initializing agents with autogen.Agent classes, binding tools like calculators or web searchers, and executing workflows in conversable_agent chats.

Semantic Kernel integration forms a core pillar, teaching kernel-based orchestration for .NET and Python devs. Notebooks guide plugging in planners, memories, and plugins to create semantic functions that chain LLM calls. A standout lesson covers agentic RAG, blending retrieval from vector stores with agent reasoning to ground hallucinations—using embeddings to fetch context, then routing to specialized agents for synthesis.

What sets this apart: modular design lets builders jump to specifics, like building a code-review agent or travel planner, without prerequisites. Each notebook runs standalone, with embedded OpenAI or Azure API keys for instant execution. Recent commits refine these for edge cases, such as error recovery in long-running agent swarms and hybrid local/ cloud deployments.

For Git builders, this matters because agents are shifting apps from reactive scripts to proactive systems. Clone the repo and run jupyter notebook to prototype: lessons enforce best practices like dynamic tool calling and state persistence via JSON-serialized memories. With over 50,000 stars signaling sustained interest, it equips teams to embed agents in CI/CD pipelines or dev tools—solving the "where to start" gap in production-grade agentic AI.

Technically rigorous yet accessible, the course bridges theory to deployable code, arming developers against the agent hype with executable knowledge.

Use Cases
  • Developers prototyping multi-agent code review workflows.
  • Builders creating RAG-enhanced customer support agents.
  • Teams orchestrating Semantic Kernel for enterprise task automation.
Similar Projects
  • microsoft/autogen - Core conversational framework; this course provides structured beginner notebooks atop it.
  • microsoft/semantic-kernel - Planner-focused orchestrator; here it's contextualized in full agent curricula.
  • langchain-ai/langgraph - Graph-based agent workflows; Microsoft's emphasizes Microsoft-stack integration and Jupyter simplicity.

More Stories

ColossalAI Speeds Up Training of Giant AI Models

Open-source framework uses advanced parallelism to cut costs on high-end GPUs like H200 and B200

hpcaitech/ColossalAI · Python · 41.4k stars

Tesseract Powers Open Source OCR for Image Text Extraction

Mature C++ engine processes 100+ languages with LSTM neural nets and legacy modes

tesseract-ocr/tesseract · C++ · 72.9k stars

Quick Hits

generative-ai Dive into Jupyter notebooks with sample code to build generative AI apps using Gemini on Google Cloud's Vertex AI. 16.3k
guidance Program precise control over large language model outputs with this intuitive guidance language. 21.3k
gradio Launch interactive machine learning web apps instantly using pure Python with Gradio. 42k
awesome-mcp-servers Browse this curated collection of MCP servers to power your multi-protocol integrations. 83.3k
generative-ai-for-beginners Master generative AI basics through 21 hands-on lessons designed for quick-start builders. 108.1k

NASA JPL Open-Sources Build-It-Yourself Mars Rover Design

Scaled-down six-wheeler uses consumer parts for robotics education and rugged-terrain research without prior expertise needed.

nasa-jpl/open-source-rover Prolog Latest: v4.0.0 9.2k stars

The JPL Open Source Rover (OSR) project delivers detailed blueprints for a six-wheeled robot mimicking NASA's Mars rovers, constructed entirely from consumer off-the-shelf (COTS) components. Launched in 2017 and now at version 4.0.0, it targets builders seeking hands-on entry into mechanical engineering, electronics, software, and robotics.

At its core, OSR solves the barrier of high-cost, proprietary hardware in robotics prototyping. For around $1,600—comparable to a TurtleBot 3 Waffle—it assembles into a robust aluminum-frame machine with 10 motors powering a top speed of 1.6 m/s (slow jog pace, motor-dependent). Mechanicals draw heavily from GoBilda parts for strength and speed, though cheaper alternatives exist at the expense of durability. The design prioritizes rugged terrain navigation, with rocker-bogie suspension echoing Perseverance and Curiosity rovers.

Version 4.0 marks the first stable release of a major redesign after over five years, detailed on its Hackaday project page. Key updates include automated parts-list generation via GitHub Actions, streamlining assembly documentation. A notable tweak swaps an endcap for a two-hole beam, refining structural integrity (PR #393). Expect ongoing docs polish post-release.

Software integration supports expansions like a head display or robotic arm, making it a flexible research platform. No prior skills required: repositories provide CAD files, wiring diagrams, and assembly guides. A community gallery showcases user builds, from basic chassis to enhanced variants.

For builders, OSR matters as a low-entry teaching tool that scales to real-world challenges. It inspires solar-system exploration enthusiasm while offering a premium, customizable alternative to fragile kits. With steady development over 7.7 years and over 9,200 GitHub stars, it sustains a niche for DIY robotics without vendor lock-in.

Technically, the COTS focus—bolts, beams, motors from accessible suppliers—enables rapid iteration. International shipping via GoBilda ensures global reach. Builders can prototype autonomy stacks, sensor fusion, or teleoperation on terrain simulating Martian regolith, bridging hobby to professional R&D.

Use Cases
  • Students prototyping rugged robots for mechanical engineering classes.
  • Hobbyists assembling expandable Mars rover clones at home.
  • Researchers testing navigation algorithms on real hardware.
Similar Projects
  • TurtleBot3 - Matches cost but prioritizes indoor SLAM over outdoor durability.
  • ROS-based rovers - Emphasize software stacks like ROS2, lacking OSR's full COTS hardware specs.
  • Wild Thumper - Cheaper tank-style chassis but less scalable for arm/display expansions.

More Stories

Robotmk Integrates Robot Tests into Checkmk Monitoring

Bridges end-to-end testing with infrastructure oversight for unified application visibility

elabit/robotmk · Rust · 58 stars

MuJoCo 3.6.0 Sharpens Contact Dynamics Simulation

Google DeepMind engine boosts performance for robotics and machine learning workflows

google-deepmind/mujoco · C++ · 12.4k stars

Quick Hits

kornia Kornia equips Python builders with geometric computer vision tools for crafting precise Spatial AI applications. 11.1k
autoware_universe Autoware Universe delivers a full C++ stack for perception, planning, and control in autonomous vehicles. 1.5k
ros-mcp-server ROS-MCP-Server links Claude and GPT models to robots via MCP and ROS for instant AI-robot control. 1.1k
webots Webots simulates robots with realistic physics in C++ for rapid prototyping and testing. 4.2k
rl PyTorch RL offers modular, primitive-first Python tools for building custom reinforcement learning pipelines. 3.3k

httpx Powers Fast, Reliable HTTP Probes for Security Pipelines

Go-based toolkit delivers multi-threaded probing with retries, handling WAFs and edge cases for pentesters and OSINT workflows.

projectdiscovery/httpx Go Latest: v1.9.0 9.7k stars

Developers building security tools now have a robust option in httpx, a Go-written HTTP toolkit from ProjectDiscovery. Launched in 2020, it tackles the core challenge of gathering reliable HTTP intelligence at scale: slow, flaky probes under high concurrency. By leveraging the retryablehttp library, httpx maintains accuracy even with thousands of threads, incorporating retries, backoffs, and smart fallbacks from HTTPS to HTTP.

At its heart, httpx processes inputs like hosts, URLs, or CIDR ranges, running a suite of configurable probes. Default checks cover essentials: URL, title, status code, content length, TLS certificate, CSP header, line/word count, location header, web server, web socket, response time, body/header hash, URL scheme, and request method. Optional probes include IP, CNAME, raw HTTP, HTTP/2, HTTP pipeline, virtual host, CDN, paths, ports, favicon/JARM hash, redirect chain, and ASN.

This modularity suits pipeline integration, with simple CLI flags for customization. Edge-case handling shines: it dodges WAF blocks via exponential backoffs and supports virtual hosting. Installation is straightforward for Go 1.25+: go install -v github.com/projectdiscovery/httpx/cmd/httpx@latest.

The v1.9.0 release, pushed in March 2026, adds firepower. New ML-based classifiers via "dit" detect page types (login, captcha, parked) with -fpt flag, deprecating older error-page filters. Database outputs now store results in MongoDB, PostgreSQL, or MySQL. Markdown export joins JSON/CSV options. Fixes include graceful shutdowns to prevent data loss on interrupts and improved resolver parsing.

With 9,695 GitHub stars reflecting steady five-year traction, httpx prioritizes CLI reliability over service deployment—active development warns of breaking changes. Pentesters value its speed for recon; OSINT builders chain it in workflows; bug bounty hunters probe live targets efficiently. Review changelogs before upgrades, as it's tuned for standalone use.

Supported Probes:

  • Defaults: URL, Title, Status Code, Content Length, TLS Certificate, CSP Header, Line/Word Count, Location Header, Web Server, Web Socket, Response Time, Body Hash, Header Hash, URL Scheme, Request Method
  • Optionals: IP, CNAME, Raw HTTP, HTTP2, Pipeline, VHost, CDN, Paths, Ports, Favicon/JARM Hash, Redirect Chain, ASN

For documentation, see ProjectDiscovery's site.

Use Cases
  • Pentesters probing URLs for live status and TLS details.
  • OSINT analysts scanning CIDR for web servers and CDNs.
  • Bug bounty hunters chaining probes in recon pipelines.
Similar Projects
  • httprobe - httpx expands with 20+ probes, retries, and WAF handling beyond basic aliveness checks.
  • whatweb - httpx emphasizes speed and concurrency for pipelines over whatweb's deeper web tech fingerprinting.
  • aquatone - httpx skips screenshots for broader, faster CLI probe suite without visual reconnaissance.

More Stories

MITRE Caldera Automates Adversary Emulation for Red Teams

Python-based platform runs MITRE ATT&CK operations via C2 server and plugins

mitre/caldera · Python · 6.8k stars

Suricata Powers Network Intrusion Detection and Prevention

Open-source engine from OISF inspects traffic for threats with multi-mode security monitoring

OISF/suricata · C · 6.1k stars

Quick Hits

caddy Caddy serves fast, extensible HTTP/1-2-3 across platforms with automatic HTTPS, ideal for builders craving effortless secure web hosting. 70.9k
ImHex ImHex delivers a retina-friendly hex editor for reverse engineers and programmers powering 3 AM deep dives into binaries. 52.9k
authelia Authelia secures web apps via certified OpenID Single Sign-On and multi-factor authentication in a sleek portal. 27.2k
cilium Cilium harnesses eBPF for fast networking, security, and observability in cloud-native infrastructures. 24k
hacktricks HackTricks aggregates CTF, real-app, and research tricks to equip security builders with battle-tested exploit techniques. 11k

Rust-Built Ruff Revolutionizes Python Linting with Extreme Speed

This drop-in replacement for Flake8 and Black lints and formats code 10-100x faster, tackling monorepo-scale challenges head-on.

astral-sh/ruff Rust Latest: 0.15.6 46.3k stars

Python developers have long grappled with slow linters and formatters that bog down workflows, especially in large codebases. Enter Ruff, an extremely fast linter and code formatter written in Rust. Launched in 2022 by Astral—the team behind uv and ty—it delivers 10-100x speedups over tools like Flake8 and Black, often completing checks in fractions of a second on massive projects.

Ruff's core strength lies in its Rust implementation, which enables parallel processing and efficient static analysis without Python's runtime overhead. It supports pyproject.toml configuration, Python 3.14 compatibility, and built-in caching to skip unchanged files. With over 800 rules, it natively reimplements popular Flake8 plugins like flake8-bugbear, plus equivalents for isort, pydocstyle, pyupgrade, and autoflake. Developers can auto-fix issues, such as removing unused imports, via --fix.

What sets Ruff apart is its drop-in parity: it replaces multiple tools under one interface, reducing dependency sprawl. Monorepo users benefit from hierarchical configs, making it ideal for sprawling projects. Editor integrations for VS Code streamline adoption, while first-party fixes ensure reliability.

Major open-source projects already rely on it: Apache Airflow, FastAPI, Pandas, SciPy, and Hugging Face. FastAPI creator Sebastián Ramírez notes its speed is so blistering he sometimes inserts bugs to verify it's running. On a 250k-line module, it clocks in at 0.4 seconds versus minutes for alternatives.

The latest release, 0.15.6 (March 2026), introduces preview features like lazy import parsing and star-unpacking for comprehensions (PEP 798). New rules flag Airflow-specific issues, such as Variable.get() outside task contexts (AIR003) and runtime-varying DAG args (AIR304). It adds delattr-with-constant (B043), enforces/bans lazy imports (TID254), auto-inserts from __future__ import annotations (UP006), and supports newline in refurb rules for Python 3.13+.

For builders maintaining Python codebases, Ruff solves the speed bottleneck, enabling frequent linting without CI delays. Install via pip install ruff and run ruff check .—it's that simple. With 46k GitHub stars signaling broad traction, it's a toolchain essential for 2026 and beyond.

(Word count: 362)

Use Cases
  • Monorepo maintainers linting millions of Python lines instantly.
  • FastAPI teams enforcing style with auto-fixes on CI.
  • Data scientists formatting Pandas codebases in seconds.
Similar Projects
  • Flake8 - Python-based linter Ruff replaces with 10-100x speed and broader rule coverage.
  • Black - Uncompromising formatter outpaced by Ruff's integrated, faster alternative.
  • isort - Import sorter fully emulated in Ruff for unified, accelerated workflows.

More Stories

OBS Studio Delivers Open-Source Streaming and Recording Power

C-based tool composites video sources for live broadcasts and local captures across platforms

obsproject/obs-studio · C · 71k stars

fzf Powers Command-Line Fuzzy Matching for Developers

Portable Go tool filters lists like files, history and processes with interactive speed

junegunn/fzf · Go · 78.7k stars

Quick Hits

prometheus Prometheus builds scalable monitoring by scraping metrics into a time series database with powerful querying and alerting. 63.2k
git Git enables distributed version control with lightning-fast branching, merging, and history tracking for team coding. 59.7k
FFmpeg FFmpeg handles all multimedia tasks like recording, converting, streaming, and transcoding audio/video formats. 58k
php-src PHP interprets server-side scripts to power dynamic websites with database integration and session management. 40k
curl cURL transfers data across 30+ protocols using its robust CLI tool and versatile library for any app. 41.2k

Embedded Engineering Roadmap Charts Path Through Hardware-Software Frontier

Curated guide equips aspiring and practicing engineers with structured resources to master embedded systems development.

m3y54m/Embedded-Engineering-Roadmap Unknown Latest: v1.2.3 9.8k stars

For builders eyeing a career in embedded systems, where hardware meets software in resource-constrained environments, the m3y54m/Embedded-Engineering-Roadmap offers a no-nonsense blueprint. This GitHub repository delivers a comprehensive learning path, blending electronics fundamentals with firmware programming—essential for systems like aircraft controls or transit controllers.

Embedded systems, per the ISO/IEC/IEEE 24765 standard, are computer systems integral to larger setups, optimized for reliability, cost, size, and power. The roadmap echoes this with definitions from sources like Making Embedded Systems ("purpose-built computerized systems") and Analog Devices ("computers invisibly embedded in products"). It underscores the field's rigor: hardware is hard, demanding motivation and hands-on projects.

Targeted at beginners aspiring to become embedded engineers/developers and practitioners seeking skill expansion, the roadmap structures progression logically. It starts with core concepts—microcontrollers, microprocessors, digital signal processors—then advances to firmware engineering, electronics, and full-system integration. Curated resources include books, tutorials, and tools, emphasizing practice over theory.

What sets it apart is its dual focus: hardware functionality (e.g., circuit design, signal processing) paired with software skills (C/C++, real-time OS like FreeRTOS). Unlike scattered online tutorials, this provides a sequential map, helping avoid common pitfalls like ignoring power optimization or reliability testing.

The project, now 2.6 years mature with steady updates, hit latest release v1.2.3 featuring minor refinements to resource links and structure. Last pushed in March 2026, it reflects ongoing maintenance amid nearly 10,000 stars from a dedicated community.

For embedded work—think IoT devices, automotive ECUs, or medical implants—this roadmap demystifies entry barriers. It warns of challenges but promises proficiency through dedication: build projects, debug peripherals, iterate on prototypes.

Builders in firmware, hardware, or software engineering will find it invaluable for upskilling. Whether prototyping on Arduino/STM32 or scaling to industrial ARM cores, it delivers the signal: patient practice turns novices into pros.

Use Cases
  • Aspiring engineers mapping beginner-to-expert embedded learning journeys.
  • Firmware developers expanding hardware-software integration skills.
  • Practitioners refreshing electronics and optimization knowledge bases.
Similar Projects
  • kamranahmedse/developer-roadmap - General software paths lack embedded hardware depth and firmware specifics.
  • mark-rodseth/embedded-roadmap - Narrower hardware focus, omits comprehensive software and project-building guidance.
  • EmbeddedArtistry/embedded-resources - Resource aggregator without structured career progression roadmap.

More Stories

Ultrasonic Jammer Blocks Hidden Microphone Recordings

ATTINY Microcontrollers Drive 25kHz Piezos to Disrupt Spy Devices and Voice Assistants

mcore1976/antispy-jammer · C++ · 353 stars

Docker Stack Unifies Self-Hosted D&D Group Tools

FoundryVTT, Discord bots, wikis and Node-RED flows enable complete tabletop RPG infrastructure

oakbrad/dungeonchurch · Shell · 87 stars

Quick Hits

micrOS Whip up async automation for DIY gadgets with micrOS, a lightweight Python OS that supercharges project control. 128
iiab Turn a Raspberry Pi into an offline Library of Alexandria with Internet-in-a-Box for instant knowledge access anywhere. 1.8k
ghdl Simulate VHDL 2008/93/87 hardware designs precisely with GHDL, streamlining your FPGA prototyping workflow. 2.8k
node-feature-discovery Auto-detect Kubernetes node hardware features for smarter cluster optimization and dynamic workloads. 1k
p3a Animate pixel art flawlessly on ESP32-P4 with p3a, ideal for compact retro display builds. 56
Memes section coming soon. Check back tomorrow!