Links
AddOpen source Web Application Firewall for APIs and web applications. Enterprise-grade protection, 100% compatible with OWASP Core Rule Set.
Related contents:
Tapez n'importe quoi, l'IA tranche. Le modèle tourne entièrement dans votre navigateur : rien n'est envoyé à un serveur.
Related contents:
A Multi-Agent Framework for Automated AI Research Paper Writing
PaperOrchestra is a multi-agent framework for automated AI research paper writing. It flexibly transforms unconstrained pre-writing materials (ideas and experimental logs) into submission-ready LaTeX manuscripts, including comprehensive literature synthesis and generated visuals, such as plots and conceptual diagrams.
A Claude Code skill bundle for bug hunting and external red-team work - 82 skills, 15 slash commands, 681 disclosed-report patterns curated across 24 core vulnerability classes, plus enterprise identity + infrastructure attack matrices.
Web browsers are ubiquitous, but how do they work? This book explains, building a basic but complete web browser, from networking to JavaScript, in a couple thousand lines of Python.
One stylesheet for the HTML your server renders
Dialogs, menus, drawers, tabs and toasts that work with JavaScript turned off. No build step. It fits Rails, Django, Laravel and htmx, and works just as well in React, Vue or one plain HTML file.
Pyodide is a port of CPython to WebAssembly/Emscripten. Pyodide is a Python distribution for the browser and Node.js based on WebAssembly.
Pyodide makes it possible to install and run Python packages in the browser with micropip. Any pure Python package with a wheel available on PyPi is supported. Many packages with C, C++, and Rust extensions have also been ported for use with Pyodide. These include many general-purpose packages such as regex, PyYAML, and cryptography, and scientific Python packages including NumPy, pandas, SciPy, Matplotlib, and scikit-learn.
Related contents:
Another Build File Service (ABFS) speeds up the checkouts and builds of large code repositories. The initial focus is on accelerating Android development but there is also some ongoing work to add Bazel support. Internally at Google, we see Android checkouts in seconds and builds in minutes.
ABFS enables lightweight code checkouts and enables builds to share a cache of previously performed build actions. If you have CI running builds, multiple developers running builds, or develop across multiple Android branches, ABFS can help you.
Related contents:
Discover and install source-backed runtime-targeted and portable skills for Codex, Claude Code, Cursor, diagrams, and repository workflows.
Public Agent Skills for Codex operations, Cursor operations, Claude operations, repo maintenance, skill maintenance, productivity, and engineering workflows.
This crawler helps to index binary documents such as PDF, Open Office, MS Office.
Related contents:
Java on NVIDIA GPUs, Apple Silicon, and more.
TornadoVM JIT-compiles JVM bytecode to CUDA, OpenCL and Metal — and now natively integrates cuBLAS, cuFFT, and cuDNN inside your Java code. Write kernels in Java, keep your build on Maven, and let the runtime generate the GPU code. No JNI glue, no second toolchain, extreme performance.
Related contents:
Autonomous penetration testing that keeps its receipts. AI autonomous penetration testing system.
Self-hosted multi-agent offensive security. A planner explores, workers run real tools, and every finding stays bound to the assets and HTTP traffic that prove it.
Related contents:
machine learning in Python.
scikit-learn is a Python module for machine learning built on top of SciPy and is distributed under the 3-Clause BSD license.
Related contents:
Android leak audit.
Desktop Android leak audit: permissions, endpoints, data flows, reversible fixes. Your phone has 500 apps. How many phone home?
NOIR reads what installed apps can collect and where they send it — permissions, endpoints, data flows — then hands you reversible fixes.
Install Linux next to Windows without a USB stick.
hop is a single Windows app that installs Linux alongside Windows. Pick a system, choose a drive, drag to choose how much space it gets, and click Install. There's no USB stick to find and nothing to install first.
244 pixel art effects, every one a small piece of code. Plus a Claude Code skill to make your own.
244 pixel art effects and animations, every one a small piece of code. Fire, explosions, magic, weather, UI, pickups, critters and more. Play them live, recolor and mirror them, drop them into a browser game, or export sprite sheets for any engine. Plus a Claude Code skill that makes new ones the way a picky pixel artist would: draft, render, look, refine.
Check your agent skills for quality and security, then measure whether they help an agent complete real tasks.
Multi-tier framework for evaluating AI agent skills and plugins with quality gates, semantic overlap detection, synthetic evaluation dataset generation, and live agent evaluation that measures how skills affect agent behavior.
stateless browser running entirely on Workers.
Kitesurf is an experimental browser engine that runs on Cloudflare Workers. It parses HTML, applies CSS, runs page JavaScript, and renders screenshots without Chrome or a browser container.
Related contents:
An open protocol for every company and every personal agent.
Personal agents can discover participating companies, sign users in, and get things done through APIs, web pages, and company agents. Companies decide how to empower agents and what they can do.
Related contents:
Tool for converting markdown files into anki decks.
Markdown anki decks uses the question to uniquely identify the card. You can change the card contents without losing your progress on the card. Markdown anki decks can be reimported without creating duplicates.
Clojure on Scheme. A Clojure compiler implemented on top of Chez Scheme.
Jolt is a self-hosted Clojure implementation that runs on a portable Scheme layer that supports multiple backends. Chez natively, Gambit for JavaScript, no JVM. Most Clojure runs unchanged, any C library binds natively, and jolt build ships a single standalone binary.
Turn a real-world benchmark into a formally verified one.
Benchproofer takes the tasks of an existing benchmark (SWE-bench, HumanEval, MBPP, QuixBugs, GSM8K, BigCodeBench, or your own file of tasks), formalizes each one into a specification plus an implementation a verifier checks, has each formalization attacked by a language model that did not write it, and then adjudicates the result mechanically. Every claim about a task is a verdict some gate produced and recorded, so anyone can re-run the gates on someone else's artifacts and get the same verdicts. Nothing certifies itself.
Related contents:
AI meeting notes that stay on your computer. Yogurt records your calls without adding a bot to the meeting.
Private AI meeting notes. Free and open source. Because your business should never be anyone else’s.
A package manager for AI agent tooling. Manage AI agent tooling once. Sync it everywhere you work
Manage skills, commands, hooks, and MCP servers from one config, then sync them to Claude, Cursor, Codex, and more.
agent-marketplace is a command-line tool that installs shared AI agent resources into the AI coding tools you use. A resource is a skill, command, subagent, hook, or MCP server. The tool reads resources from registries (Git repositories or local folders) and from hosted Marketplaces, then writes each one in the native layout of every harness you target: Factory Droid, Claude Code, Cursor, Windsurf, Gemini CLI, Devin CLI, Codex, and OpenCode.
Unofficial agent skills for Google SRE practices (SLOs, incidents, AI ops). Not affiliated with Google.
55 tactical Cursor Agent skills that turn Google’s SRE literature into workflows you can run in chat: design SLIs/SLOs, cut pager noise, manage incidents, review resilience, ship canaries, adopt AI ops safely, and more.
One DB client to replace them all. a fast, native database client.
Gets out of the way of you and your data. GPU Accelerated, no Electron, no JVM, no Tauri, fast, memory efficient and snappy on datasets where others crash.
The modern container runtime. 8× faster than Docker. a fast Docker Desktop alternative for Mac, Windows, Linux
Kiri runs Docker containers with one daemon. The docker CLI, Compose and the Docker API work as they do today.
Kiri is a lightweight replacement for Docker Desktop and OrbStack, written in Rust. It runs containers in one small Linux VM (or straight on the kernel on Linux) and speaks the Docker Engine API, so your tools just work.
Linux Server Management Panel for VPS Servers. Run your servers without the terminal.
Deploys, backups and fixes for the VPS you already rent. Connect it with one command.
Kanister Backup & Restore
Recovery Contract for Kubernetes: backups are restored, verified against measurable objectives, and marked PROVED or UNPROVED before an incident — not assumed after pg_dump succeeds.
Related contents:
Model capability is not research capability. A capable model can still produce weak vulnerability research when it lacks target context, deterministic tools, authorization boundaries, durable state, and a validation method.
In this workshop, you will compare a generic agent response with an investigation performed through a target-specific harness. You will preserve a baseline, select and scope an authorized open-source target, build reusable research context and procedures, and test one specific security hypothesis when the target can run safely.
Open-source detection signatures for AI/ML attacks and breaches.
A vendor-neutral collection of Sigma, YARA, and Suricata rules for detecting compromise of LLM applications, MCP servers, ML supply chains, AI infrastructure, AI-powered insider threats, and RAG/vector database attacks.
A curated list of red teaming tools, C2 frameworks, and adversary simulation resources - organized by kill chain phase for offensive security professionals, penetration testers, and purple teamers.
Qwen’s agentic coding tool that lives in your terminal and helps you turn ideas into code faster than ever before.
Run Rust code without the wait.
rustc builds your program, then you run it. Relax runs your program while it builds it.
Relax is a live compiler and JIT runtime for Rust, written in Rust. It runs a Cargo package straight from source, with no rustc, LLVM or linker in the loop: it compiles each function when it is first called, optimises the hot ones while the program runs, and patches your edits into the running process.
Free, open-source embroidery software for Mac and Windows. Turn SVG artwork into machine embroidery, edit stitches visually, and create designs with AI agents via MCP.
ZigZag is a visual embroidery editor built in Rust for performance, with built-in MCP (Model Context Protocol) support so AI agents can create and edit designs alongside you. Turn SVG artwork into stitches, fine-tune it in the editor, and export PES, DST, EXP, JEF, or VP3 for broad compatibility with embroidery machines—or send designs directly to a Brother Skitch (PP1) over Bluetooth.
Make working with many Kubernetes clusters super easy regardless of where they are deployed
Open Cluster Management is a community-driven project focused on multicluster and multicloud scenarios for Kubernetes apps. Open APIs are evolving within this project for cluster registration, work distribution, dynamic placement of policies and workloads, and much more.
If you like Open Cluster Management, give it a star on GitHub!
Related contents:
Run LLMs, Fast.
Run open models on the GPUs you own, with native Metal and CUDA inference.
TensorFold serves language models from a Zig binary on Apple Silicon and NVIDIA GPUs. The engine reads checkpoints, tokenizes requests and runs Metal or CUDA kernels directly. Serving needs no Python or MLX installation.
AI coworkers. Ready for the enterprise.
OpenBot is the modular, self-hostable, enterprise-ready agent workspace with a clean architecture and built-in governance. Bring the agents your engineers build or start with ours, then choose the models, tools, and infrastructure they run on. Use it as-is, white-label it, or deeply customize it.
Create Professional Demos and Edit Videos in Seconds.
Record your screen or upload a video, add smooth zooms, device mockups, 3D effects, and custom backgrounds - export a cinematic demo.
Free Accessibility Checker & WCAG 2.2 Audit Tool.
A zero-dependency WCAG 2.2 accessibility audit engine, written from the W3C specifications. Runs entirely in the browser and reports what it cannot prove instead of guessing.
The OpenRC init system.
OpenRC is a dependency-based init system that works with the system-provided init program, normally /sbin/init.
AWS Emulator is a local emulator of the AWS APIs.
AWS Emulator is a local emulator of the AWS APIs. You use it to test your software without calling AWS, on your own machine, on a development server, or in continuous integration (CI). You don’t need an AWS account. Your requests go to the emulator, not to AWS. So you work with emulated resources instead of AWS resources. AWS Emulator has no additional service fee. You pay only for AWS resources when you deploy on AWS.
Use your most capable model to audit your codebase and write plans for cheaper models to execute.
An agent skill that audits any codebase and writes implementation plans for other agents to execute.
The idea: use your most capable model for the part where intelligence compounds — understanding the codebase, judging what's worth doing, writing the spec — and hand execution to cheaper models. The skill never implements anything itself. The plan is the product.
Hyper-Realistic Sand Simulator in Your Browser. Hyper-realistic sand on your phone screen (WebGL2).
Real-feeling sand on your phone screen. Tilt the phone and it slides, jerk it and it lags behind, drag a finger and it plows a channel, flip the phone face-down and it falls off.
Sandboxes for Coding Agents.
Docker Sandboxes run AI coding agents in isolated environments on your machine or on Docker-managed cloud infrastructure. Use the sbx CLI to create and manage either kind of sandbox.
Related contents:
Most people learn React through hooks. useState, useEffect, useMemo, and a list of rules to follow without really knowing why.
But here’s the thing, React is not about hooks. Hooks are only the surface, the part you touch. So forget about them for now.
Underneath, React is a scheduler. It decides what to render, when to render it, and what can wait. It talks to the browser, splits work into small pieces, and pauses when something more important comes in.
A context-compression plugin for month-long single sessions (billions of tokens) and small context windows (a 100K context is enough), billion-context is all you need .
60-120fps / 2160p tab video/audio capture from Chrome.
Record a Chrome tab at 60-120fps (up to 2160p) straight to a AV1 .mp4. Tab audio too. No ffmpeg.
A Model Context Protocol server that enables AI assistants to interact with Home Assistant. Control devices, query states, manage automations, and configure dashboards using natural language.
Actual cost of Argo CD applications, from OpenCost, inside the Argo CD UI.
Related contents:
Give AI agents human-like PDF vision. Extract PDF signals for AI agents.
Turn silent failures in PDF extraction into errors agents can recover from. On its own, extraction reports empty scans, glyph garbage, and scrambled columns as success — pdfvision flags them page by page and names the next step: render the page, run OCR, or crop to the region that matters.
CLI, MCP server, and TypeScript library that extracts text, layout, and page images from PDFs for AI agents — when extraction goes wrong, it says so per page and says what to do next.
Mermaid diagrams, drawn in your terminal.
Mermaid diagrams drawn as ASCII in your terminal and in Claude Code. You can also view markdown files with mermaid code blocks.
A dependency manager for AI agents -- like npm for agent context.
APM is a dependency manager for AI agents. Declare the skills, prompts, instructions, plugins, and MCP servers your project needs in one apm.yml, then any developer runs apm install to deploy each primitive to the harnesses that support it. See the targets matrix for support across GitHub Copilot, Claude Code, Grok Build, Cursor, OpenCode, Codex, Gemini, Windsurf, and Kiro.
Securely synchronize files on-demand using Syncthing on iOS and macOS.
Related contents:
A translation-based Apple Silicon layer for x86-64 PCs.
A translation-based Apple Silicon layer for x86-64 PCs. It is a UEFI boot program with an arm64-to-x86-64 translator, a Metal driver for NVIDIA GPUs and a model of Apple silicon hardware. Stela currently presents itself as an M2 Mac mini.
Related contents:
Viz. Apple IntelligencePlatform store.
macOS builds a knowledge graph about the person who uses it: who you know, how it decided those people are related to you, who you write to and how often, whose faces recur in your photos, where you might be next. Check out the write up.
Apple documents very little of this. The daemon that builds the graph gets a one-line man page, and there is no user-facing account of what it stores or what it infers. iKnow is a look under the hood: a small web app that opens the graph on your own Mac and shows you what is in it, what was recorded, what was inferred, and which other parts of the system read it.
Related contents:
Tell agents, AI crawlers, search crawlers and bots apart from people. agent-traffic classifies HTTP requests. It’s a curated, tested dataset of the user agents, headers and official verification sources of real agents and crawlers, plus one small function that applies it.
A boot animation showing stages of systemd as a tree.
A live tree of service manager jobs on the console, for systemd and OpenRC, in the spirit of nix-output-monitor. It replaces the vertically scrolling kernel and service lines with the units still pending. In the room the tree leaves, a timeline is displayed of the slowest ones that ran.