3D Model Anything is an open-source iPhone LiDAR capture app, Three.js viewer, and agent skill for reconstructing rooms as editable Three.js scenes. The app records room geometry and photographs; the scan-to-threejs skill guides an AI coding agent through measurement, modeling, and matched-camera visual comparisons rather than performing a one-click conversion. The repository includes an editable office demo, browser views such as dollhouse, walking, overhead, wireframe, and knolling modes, and workflows for importing user captures. LiDAR-equipped iPhones or iPads are required for scanning, while non-LiDAR devices can import and review scans. The reusable source code and skill are MIT licensed; the example's images and textures have separate terms, and the repository does not distribute other rooms or original scan bundles.
Anatomy is an open-source agent skill for Claude Code, Antigravity, OpenCode, and Codex that explains technical ideas by constructing interactive isometric illustrations of physical machines, instruments, tools, or other mechanisms. It uses a workflow that collects the relevant numbers, invents a physical object for the concept, creates a parts list and geometry, renders parts back to front, adds interaction, and verifies the result with geometry and browser checks. The skill can produce standalone HTML pages, embedded SVG files, or React/Next.js components. Its SVG kit provides isometric solids, round parts, pipes, cables, four-tone shading, animation, cards, and live readouts; optional WebGL layers add effects such as fire, exhaust, water, caustics, light, and glow, while a beta 3D mode re-projects and re-shades moving parts each frame. The repository includes Node build scripts, complete examples, React components, reference guides, and headless Chrome tools for screenshots, scripted interaction, line-end checks, and geometry audits. It uses Node.js built-in modules without npm dependencies for the kit and examples, with React 18+ required for React output and Chrome or Chromium required for browser checks. The project is distributed under the MIT license. Its stated limits include not being a chart library, not targeting people or exact maps, and requiring WebGL for shader effects.
BRUTAL is an open-source C# framework for making games or building a separate game engine. It provides a custom interop layer for direct access to native APIs, along with low-level libraries for unmanaged memory, collections, strings, mathematics, concurrency, logging, localization, and bindings for graphics, windowing, audio, networking, physics, and other native components. BRUTAL deliberately has no editor, scene graph, renderer, or general-purpose engine architecture. Developers assemble their own stack from separate NuGet packages, such as packages for Vulkan, GLFW, Dear ImGui, FMOD, RakNet, Jolt, and Box2D; the framework uses unmanaged memory, unsafe code, and allocation-avoiding components where appropriate. The repository contains the source for its published packages. BRUTAL is developed by Ahwoo Limited and is free to use, modify, and distribute, including in commercial games, under a licence based on MIT with attribution requirements.
CCBaga is a native desktop toolkit for managing local Claude Code and Claude Desktop identities and verifying the surrounding computing environment. It archives the current local identity before switching to a saved or empty workspace, lets users choose whether skills, commands, plugins, MCP configuration, settings, project records, conversation history, and caches move with the identity, and stores archives in authenticated encrypted .baga containers tracked by a local SQLite database. The project also includes CCBaga-B, an isolated Fingerprint Chromium companion browser: CCBaga-A applies a selected region's language, timezone, and system settings, generates a browser fingerprint seed, restarts the browser, and compares browser and machine readings with the network exit through a local session. An optional separate Network Skill follows the Agent Skills specification and uses an AI coding agent, OpenSSH, Bash, and Xray to configure an authorized Debian server and client after explicit approval. The toolkit includes native workflows for detecting, uninstalling, or reinstalling Claude Code through supported package managers. It runs on Apple Silicon macOS 12 or later, x86_64 Linux with GTK 3 and WebKitGTK 4.1, and Windows 10/11 x86_64; Intel macOS and Linux ARM64 are not supported. CCBaga's original code is released under the MIT License, while third-party components retain their own terms, and the project states that it is independent of Anthropic and Apple.
CleanYourTV is an open-source toolkit for debloating Android TV and Google TV devices over Wi-Fi through ADB and installing a native Apple TV-style home launcher. It backs up installed packages and Home-screen state, disables selected manufacturer telemetry, advertising services, content rows, and preinstalled packages with `pm disable-user`, and installs the launcher without root access, firmware changes, or uninstalling packages. Its restore and recovery scripts reverse the changes, while model-specific debloat lists and discovery prompts account for differences between TV brands and firmware. The repository also includes a pure-Python ADB client, launcher build and installation scripts, privacy auditing, theme support, firmware-drift checks, screenshots, launcher benchmarking, and performance tools, with Windows PowerShell counterparts. The launcher supports Android 8.0/API 26 or newer. Fire TV supports only the debloating portion; Samsung Tizen and LG webOS are different platforms. The project is distributed under the MIT license.
Day Trader skill is a free, MIT-licensed set of 11 Claude skills for practicing US-stock day trading with plain Python and free public data. Its skills pass shared files through a workflow: chart scanning, news monitoring, entry timing, exit planning, replay-based trade analysis, position sizing, risk checks, broker tickets, alerts, earnings analysis, and congressional-trade tracking. The Time Machine replays possible outcomes from a stock's recent intraday bars and prior occurrences of the same setup; the project presents this as a simulation rather than a prediction. Paper trading is enabled by default, and the broker skill requires approval for an exact order after the risk check; the repository says it cannot place an order autonomously. The tools require Python 3.8 or newer, use standard-library Python without package installation, and read sources including Yahoo Finance, Nasdaq Trader, SEC EDGAR, and House Clerk data.
DeepInfer is DeepSeek's open-source, high-performance inference framework for offline and large-scale inference of DeepSeek models. It provides optimized kernels and model and sampler runtimes, attention data and context parallelism, mixture-of-experts expert parallelism with EPLB, speculative decoding through DSpark and confidence-scheduled verification, bounded replay for decoder prefill, asynchronous CPU batch scheduling, batch-invariant inference, deterministic sampling, and structured outputs constrained by regular expressions, JSON Schema, or EBNF grammars. The repository currently documents support for DeepSeek-V4.1-Flash on NVIDIA SM100 GPUs and Huawei Ascend 950 NPUs. It runs on Linux with Python 3.12 and requires CUDA 13 for NVIDIA or CANN 9.2 for Ascend. The release does not include a KV-cache system, prefill/decode disaggregation, or load balancing across data-parallel replicas needed for a complete inference service. DeepInfer is released under the Apache License 2.0.
DigUp is a free, open-source macOS menu bar app for searching files by their contents, including images, PDFs, documents, audio, video, screenshots, and optionally source code. It runs on Apple Silicon Macs with macOS 14 or later and keeps indexed data on the device. DigUp uses Google DeepMind's multimodal EmbeddingGemma 2 through llama.cpp and Metal. It converts images, document passages, video frames, and overlapping audio segments into a shared vector space, then combines semantic nearest-vector search with exact-word matching. Results can open a video or recording at the relevant moment, a PDF at the matching page, or code at the relevant line in a configured editor. The app reads only selected folders, excludes sensitive and ignored file types, and does not upload indexed files. It downloads the model once and checks for updates, with telemetry and accounts not required. DigUp can be installed from a DMG or Homebrew, includes a command-line tool, and is licensed under the MIT License.
DS Bot is a community plugin and desktop application for DeepSeek Harness (DSH) that lets users create and work with teams of long-lived AI bots. Each bot can have its own model selection and memory; a main bot called Chief manages other bots and group chats, while bots can hand work to one another through @ mentions and shared conversations. Group chats have separate memory and an administrator bot that coordinates responses. A shared capabilities library provides MCP servers, skills, and packs that can be assigned to bots. The bots can also create, inspect, and edit Office documents, including DOCX, PPTX, and XLSX files, subject to the runtime available in the desktop or web deployment. DS Bot is distributed as desktop applications, a DSH plugin, or a Docker deployment, and stores its data locally in the DSH home directory. It is an independent community project licensed under Apache License 2.0 and is not an official DeepSeek product.
earshot is an open-source audio analysis and sound-generation tool for coding agents. It runs locally on CPU through a CLI, Python API, or MCP server and turns audio into evidence an agent can interpret, including calibrated sound descriptions, vocabulary matches, candidate rankings, moments found inside long recordings, real-versus-code-synthesized assessments, listening timelines, measurements, annotated images, and video sound-sync critiques. Its forge component generates effects from physical-synthesis recipes such as modal resonators, filtered noise, room and microphone models, and layered DSP, then searches and refines variants using audio descriptions, reference recordings, spectrograms, model-free measurements, and artifact checks. The fit component analyzes a recording by synthesis and writes a parameterized recipe that can render variations or morphs. A bundled sound-effect bank can render coded effects by identifier, while kit commands assemble deterministic film sound kits with measured onset, peak, and end points. The project includes an MCP server for Claude Code and other agents, static HTML galleries for generated sounds, and optional audio-LLM critique with explicit consent and a user-supplied key. Its code is MIT-licensed; the two open models it uses have separate licenses. Installation requires Python 3.10+ and ffmpeg, and the first model use downloads approximately 1 GB of model files for local caching.
ESP32 iPhone Agent is an open-source AI automation tool that lets a vision model control a real iPhone without installing an app, using a computer, AirPlay mirroring, and an ESP32-S3 board acting as a USB mouse and keyboard. UxPlay receives the AirPlay stream, a GStreamer pipeline converts it into JPEG frames, and the computer sends relative pointer, typing, scrolling, and key commands to the board over Wi-Fi. A planner model creates a short plan from the task and current screen, while an actor model examines each subsequent screenshot and selects actions such as tapping, swiping, drawing, typing, pressing keys, going home, or finishing. The server tracks and recalibrates the relative pointer, checks whether the screen changed after each action, and can retry or re-plan after missed actions, repeated actions, dropped links, or invalid model responses. The project provides a Python/FastAPI server and browser-based interface with live phone control, logs, screenshots with action overlays, manual controls, calibration, and an emergency stop. It runs on macOS or Linux with an OpenAI-compatible vision model, either remote or local; the phone requires AssistiveTouch but no Mac, developer account, jailbreak, or installed app. The repository is licensed under MIT.
The Everything Project is an open-source monorepository that rebuilds common proprietary software as free alternatives, covering office, collaboration, creative, engineering, productivity, finance, business, developer tools, and games. Its applications include spreadsheets, documents, presentations, meetings, team chat, photo and video studios, CAD, circuit design, CRM, accounting, an API client, and games such as chess and Sudoku. Applications are developed within the repository and run from an individual app directory with Node.js 22.12 or newer; the repository states that its contents are licensed under MIT unless otherwise noted, while third-party libraries may have separate licenses.
Free My TV is an open-source, ad-free replacement home screen and screensaver for Google TV and Android TV, LG webOS TVs, and Samsung Tizen TVs. It puts selected apps on a configurable shelf and provides non-sponsored trending picks from installed services, weather, sunset-based light and dark themes, market prices, and a photo screensaver. The project can be installed through ADB on Android TV, Developer Mode on LG, or TizenBrew and Samsung Developer Mode on Tizen; it does not require root or uninstall the factory home screen, and its Original option restores the previous setup. On Android, the plain Java app builds without Gradle or libraries into an APK and uses the TV's local ADB interface to set the home activity, screensaver, and selected package states. A separate watchdog restores the stock home screen after repeated crashes. The LG and Samsung versions use the same interface in HTML, CSS, and JavaScript, packaged as webOS and Tizen applications. Content comes from TMDB, Open-Meteo, and optionally Binance and Watchmode, depending on the user's configuration. The repository is MIT-licensed and provides build and installation scripts, configuration for city, language, services, apps, weather, and prices, and optional transfer of personal photos to supported Android TVs. It states that the app has no accounts, analytics, or tracking.
gitai is an open-source, MIT-licensed version-control system designed for AI coding agents. It moves repository operations onto a server: clients fetch individual objects and sparse trees, while the server handles history queries, blame, path history, ancestry, code search, and derived-data caching. Changes land by rebasing onto the current branch tip and using compare-and-swap rather than a merge queue; concurrent lands can be batched, and accepted reference updates are recorded in an ordered changeset log. Conflicts can be resolved through a versioned, vendor-neutral protocol using an LLM provider, a deterministic tool, or no resolver. The server uses normal Git repositories underneath, with SQLite or PostgreSQL for changesets and references and file or cloud object storage for content-addressed packs. It includes a graft CLI and Go SDK, optional cache daemons, agent tokens and branch policies, semantic review rules, post-landing verification and reversion, and agents that monitor changesets. The repository describes it as an independent implementation inspired by Anthropic's internally developed Scion design, not affiliated with or endorsed by Anthropic.
gpu-unlock is a Linux kernel module that removes the GeForce tensor-core throughput cap for supported NVIDIA RTX 20, 30, and 40 series cards, increasing the available FP16, BF16, and TF32 throughput for workloads such as LLM prefill, fine-tuning, and image generation. It loads from the initramfs at boot, before the unmodified NVIDIA driver binds to the card, writes the full-speed selectors, and performs a function-level reset; it does not require a patched driver or a VBIOS flash. RTX 30 support is described as stable, while RTX 20 and RTX 40 support is experimental; RTX 50 cards are not supported. The project is Linux-only, supports opt-in unlocking for all supported cards or selected PCI addresses, and can reapply the unlock after suspend. The kernel module and local packaging are licensed GPL-2.0-only, while generated NVIDIA firmware remains under its own terms.
Hairline is a dependency-free, ESM JavaScript library of interactive isometric line figures for React and other DOM environments. Its figures respond to pointer movement, including pillars that rise around the pointer, layered objects that separate, cards that stand up, and other scenes such as terminals, keyboards, routers, and commit graphs. The package provides React components and DOM functions, supports Svelte-style actions, and lets each figure accept intensity, theme, accessible label, and caption callbacks. Figures draw at a 5:4 aspect ratio; the library includes server-rendering behavior, reduced-motion handling, shared animation scheduling, CSS custom properties for theming, and no runtime dependencies. It is distributed through npm under the MIT license. Hairline also provides hairline-create, a skill for coding agents that generates new figures as standalone HTML files according to the project's rules.
Harness² is an open-source framework for recursively optimizing the system around an AI model while keeping the model weights fixed. At the task level, it uses a contrastive Propose–Probe–Compose process: it proposes edits to an execution harness, probes their effects on the same task, and composes a refined harness from behavioral differences between executions without ground-truth feedback. A second recursion level maintains an improvement harness containing adopted harnesses and evidence-supported lessons for future tasks. Its editable components include system prompts, guardrails, memory, tool guidance, subagents, skills, scripts, and plugins. The repository supports parallel and sequential recursion, OpenCode and Codex executors, and benchmarks including JobBench, LAB, and WorkBuddy-Bench. It runs locally with Linux, Python 3.12, and uv, and is licensed under Apache 2.0; the repository states that it is not an officially supported Google product.
HatchZip is a Rust archive extractor with desktop and command-line interfaces that extracts archives in place while reclaiming disk space. It saves the archive index separately, extracts independent pieces from the end of ZIP, 7z, tar, and RAR archives, and truncates or deletes each completed portion so the freed space can hold subsequent output; split archives are consumed volume by volume, while tar.gz streams are processed from the front with filesystem hole punching where supported. Before changing anything, it checks the required space, preserves a configurable free-space margin, validates paths and archive integrity, and writes and flushes each output file before removing its corresponding archive data. Interrupted or crashed extractions can resume, and unfinished extractions can restore the remaining archive. It supports ZIP, 7z, tar, tar.gz, RAR volume sets, and split archives, with optional archive preservation through a normal extraction mode. HatchZip is distributed for Windows, macOS, and Linux, and is dual-licensed under Apache-2.0 or MIT; RAR support uses the separately licensed UnRAR source.
Horizon is an open-source, unofficial reimplementation of Minecraft: Java Edition written from scratch in Rust. It is distributed through the Horizon launcher and uses files from the user's legally obtained Minecraft copy; it reads vanilla worlds, follows vanilla rules, and generates the same terrain, with behavior checked against the Java game through differential harnesses. The game supports single-player worlds, vanilla-compatible saves, resource packs, data packs, multiplayer connections, dedicated servers, and Horizon's own native Rust or WebAssembly mods. Its launcher can manage installations, servers, modpacks, and updates. The accompanying `horizon-ctl` command-line tool exposes game state and player actions as JSON, while `horizon-ctl mcp` provides the same commands through an MCP server for scripts and AI agents. Horizon is available for Linux x86-64, Windows x64, and Apple Silicon macOS. It contains no Minecraft code, assets, or data and requires a Microsoft account that owns Minecraft: Java Edition. The project's original work is licensed under the MIT license; Minecraft remains subject to its owners' licensing and EULA.
leanrea is an 8 KB skill file for coding agents that guides reverse engineering of compiled applications using ordinary command-line tools instead of an MCP server. It routes targets such as Electron and ASAR files, Mach-O binaries, native disassembly, Go binaries, Android APKs, WebAssembly, SQLite databases, network captures, firmware, and web applications to appropriate first commands; its procedures use strings, imports, extracted files, and call sequences to narrow the investigation. The skill also instructs agents to work on copies, batch shell probes, search hidden extracted directories, and label claims as seen, inferred, or unknown. It can be installed for supported agents with the skills CLI or manually by copying its plain-text SKILL.md file, and is MIT licensed. The repository reports a benchmark in which leanrea matched REA's accuracy with lower wall time, cost, turns, and token usage, while noting that the comparison used one model, one machine, and three repetitions per cell.
ManimGX is an open-source Python animation engine by Academa for producing mathematical and 3D videos through an API compatible with Manim Community Edition. Its Rust and wgpu renderer exports scenes to MP4, while the same system can run in a browser using Pyodide in a Web Worker and WebGPU. It supports 2D and 3D scenes, surfaces, meshes, moving cameras, text, equations, plots, LaTeX or Typst input, synchronized speech through self.say(), and an inspect command for storyboards and 2D layout checks intended to provide feedback to coding agents. It can be installed with pip on macOS, Linux, or Windows with Python 3.13+, or used from npm for browser playback. ManimGX's own code is MIT-licensed; native wheels include components with separate licensing terms, including GPL-licensed x264.
An open-source SDK and firmware project for building custom Muse gadgets from off-the-shelf ESP32 boards, Raspberry Pi devices, or other Linux boxes. The ESP32 and Linux device SDKs let developers connect displays, buttons, audio, sensors, actuators, and custom commands, then pair the gadgets with the Muse app on iOS or Android. Gadgets require an SDK token and developer mode in the Muse app before pairing. The project is licensed under Apache License 2.0, with stated exceptions for specified third-party files and dependencies that retain their upstream licenses.
net-merge is a free, MIT-licensed network-connection aggregation tool that combines Wi-Fi, Ethernet, mobile data, and phone tethering through a virtual network adapter. It distributes an application's parallel TCP or UDP connections across available links, assigning more connections to faster links and removing unavailable links until they recover; it works system-wide without an account, relay server, VPN, or proxy configuration. It supports Windows, macOS, Linux, and Android, with command-line controls and a tray or menu-bar interface for viewing link speeds, pausing aggregation, setting per-link caps, and configuring autostart. A single connection remains limited to one link, so the main speed gains apply to applications that use multiple connections, such as download managers, browsers with parallel downloading, torrents, and updates.
OpenFilm is an open-source AI video agent and editor that lets coding agents such as Codex and Claude Code create films as HTML-based web pages. A film is a folder containing a film.html whose sections define timeline tracks of web pages, video, images, and audio; each animated page renders a frame at time t through a frame(t) function using browser technologies such as CSS, SVG, Canvas, WebGL, Three.js, GSAP, or video. Its browser-based Studio lets users move, scale, rotate, restyle, trim, and cut elements and clips, while saving edits back to the film files so agents can preserve them. The CLI can open films, inspect frames or ranges with look, render MP4 and other outputs, and obtain media through configured external providers. The look command checks frames in shuffled orders and reports changes, page errors, missing fonts or media, and text clipped by the frame. The CLI, Studio, desktop app, and format are MIT-licensed; local use requires Node, FFmpeg, and Git, while the desktop app bundles FFmpeg and a headless browser.
Owly is a Linux tool for transferring files between iPhones and Linux computers over Apple's AirDrop protocol, using supported non-Apple Wi-Fi hardware. It provides a command-line interface for sending, receiving, browsing nearby AirDrop receivers, checking hardware compatibility, and testing new cards; an optional desktop integration is available through Omdrop for Omarchy. Its owlyd system service takes control of the Wi-Fi card, runs Apple's AWDL protocol, and exposes the resulting awdl0 network interface to the user-level AirDrop application. Depending on the hardware, it uses shared mode while preserving the existing Wi-Fi connection, exclusive mode with Wi-Fi temporarily down, or active-monitor mode. The repository documents tested Wi-Fi chipsets, driver-specific behavior, channel requirements, configuration options, and known issues, and includes Rust crates for AWDL, AirDrop, and Linux radio support. Owly is distributed from its GitHub repository and through packages for Arch, NixOS, Debian/Ubuntu, and Fedora, as well as from source. Supported hardware varies by Wi-Fi card and driver; untested cards require explicit opt-in.
OXR (Optical Everything Recognition) is an open-source AI document-parsing model and service from the SoMark team. It converts PDFs and document images into agent-oriented Markdown or JSON, extracting text, tables, formulas, code blocks, chemical equations, chemical structures, and other document elements with layout bounding-box coordinates. Its architecture separates layout analysis from the OXR recognition model, while integrating reading-order handling into recognition. The layout-analysis component is independently replaceable, allowing a custom layout model to be used while retaining OXR's content recognition and reading-order capabilities. During post-training, it uses a recursive self-improvement loop that targets samples the model handles incorrectly or weakly without human annotation. OXR can run locally on Linux with a supported NVIDIA GPU and is distributed with Docker and source-based deployment options. It also provides a web demo and parsing API, with compatibility for SoMark's Python and JavaScript SDKs and CLI tools. The repository is released under the custom OXR Open Source License, based on Apache 2.0 with additional terms.
Pele is a small, synchronous open-source JavaScript library that renders Mermaid diagrams to SVG without runtime dependencies. It supports every diagram type built into Mermaid 12.1.0, exposes render() for generating SVG and mount() for rendering into a browser element, and can load individual diagram types or use lazy rendering to reduce the initial bundle. It uses CSS variables for colors and fonts, escapes diagram text, filters links and styles, and limits source and output size when handling untrusted diagrams. Pele is distributed through npm and a CDN under the MIT license.
replica-skill is a set of eleven MIT-licensed Claude skills for researching and rebuilding an existing app. The skills reverse-engineer public pages, screenshots, store listings and a user's own account into screens, flows, components and an inferred data model; plan the stack, database schema and API; reconstruct the design system and application; add authentication, payments and integrations; test flows with Playwright; compare the result with the original; analyze linked user reviews for complaints; rebrand the clone; create launch materials; and deploy it. The skills pass work between stages through files in a project's replica/ directory and can be installed in Claude Code as a plugin, copied into Claude's skills directory, or run by pasting a SKILL.md into a chat. Included Python utilities use the standard library and cover screenshot diffing, feature-parity scoring, review analysis, contrast checking, codebase-name cleanup and store-listing checks. The project states that it rebuilds functionality and user flows rather than source code, proprietary assets, branding, copy, content or private APIs, and that it uses public sources and the user's own account without bypassing logins, paywalls or site terms. It requires Python 3.8 or newer, has no package-install requirement, and is distributed under the MIT license.
Tendedero is a free, open-source native macOS screenshot utility by Alejandro Buján. It replaces the standard floating screenshot thumbnail with a line along the top of the screen where each new capture hangs. The line appears when the pointer rests in the menu bar and supports clicking to copy, pressing and holding to open Markup, double-clicking to open Preview, dragging into apps or folders, and discarding captures. On first launch, it can take over macOS screenshot settings, save captures in its own folder, and restore the previous settings when it quits or the option is disabled. It runs locally without an account, network access, or analytics, supports macOS 14 and later on Apple silicon and Intel, and is licensed under the MIT license except for the name and icon.
tsrs is a Rust-based drop-in replacement for the TypeScript 7 compiler and language server. It aims for full TypeScript conformance while using less memory and checking types faster, and can be run through its CLI with pnpm or used through Node APIs. The project also provides WebAssembly support through @tsrs/wasm for WASI, Node, and browser environments. It is licensed under Apache-2.0 and is not affiliated with or endorsed by Microsoft.
tstack is a collection of agent skills for Claude Code and Codex. Its skills automate software-development workflows such as proving, reviewing, filing, monitoring, rebasing, and merging pull requests; the council skill obtains reviews from three models from different providers and loops until they agree, while sol-loop repeatedly reviews a change with GPT Sol. Skills can call other skills, and the repository includes workflows such as full-send, full-stop, takeover, babysit-pr, modernize, file-pr, and automatic pull-request comment formatting. It is installed by copying the repository's skills into the Claude Code or Codex skills directory. An authenticated GitHub CLI and Python 3 are required for some workflows, while council and sol-loop require access to models from other providers. The repository is licensed under MIT.
ub-mode is an agent skill that applies disciplined software-engineering practices when an AI coding agent writes or changes code. Its workflow separates making code work from making it right: it uses foundational thinking, test-driven development, premise checking, domain and boundary modeling, type-system discipline, idempotent operations, root-cause fixes, behavior-focused testing, and refactoring, with work organized into verifiable units and small ordered commits. It pairs these rules with deterministic quality gates: crapper measures complexity, coverage, and CRAP scores; dryer detects structural duplication; mutator performs mutation testing; and uml-viewer visualizes components, coverage, mutation scores, and dependency directions. The skill is installed by copying SKILL.md into an agent's skills directory and invoking /ub-mode.
WordCraft is an open-source word processor for writing and document design, developed by the ArtCraft team in pure Rust as a clean-room reimplementation of Microsoft Word. It uses a ribbon-based interface with styles, tables, track changes, comments, references, mail merge, spelling and grammar tools, and document layout features. It reads and writes DOCX, imports legacy DOC and formats including ODT, RTF, HTML, Markdown, LaTeX and TXT, and exports PDF with selectable text, links and bookmarks. The application uses a shared Rust codebase with native GPU-rendered egui front ends for macOS, Windows, Linux and BSD, plus a WebAssembly browser build. Its command-driven architecture exposes the same document operations through the desktop interface, keyboard shortcuts, a command-line tool, JSON control channel and MCP server; the CLI can convert and inspect documents or run scripted editing commands. WordCraft is in alpha and is dual-licensed under MIT or Apache-2.0.
Zero Breach Toolkit is a free, open-source Python scanner for traces of known mercenary spyware, including Pegasus, Predator, Graphite and Candiru. It checks Windows, macOS and Linux computers, AndroidQF collections, and local iPhone or iPad evidence such as encrypted backups, extracted file systems and sysdiagnose archives. The toolkit compares browser, application, system, startup, network, log, file and phone records with published indicator lists. It is read-only with respect to the scanned evidence and system settings, produces readable and JSON reports with evidence grades and coverage gaps, and does not make network requests during ordinary scans. Indicator updates and optional Android collection are separate operations. A clean result is not presented as proof that a device is safe because the scanner only detects published indicators and cannot check every artifact or identify new or unpublished spyware. The project runs from source with Python 3.8 or newer and does not require administrator access for standard checks. Its code is licensed under Apache License 2.0; curated indicator files are released under CC0 1.0, while externally downloaded indicator lists retain their publishers' terms.
Searchable transcript of GitHub Trending monthly #12（2026.10） — Github Awesome (15:07). Search for a phrase, then click its timestamp to jump straight to that moment in the video.
Captions sourced from the original video on YouTube, published by Github Awesome. The video, its captions and all related intellectual property remain the property of their respective owners; AINotes claims no ownership. Provided for research, accessibility and search — see the Transcript Notice and Copyright Policy.
00:00 Here are 35 projects that caught developers attention in October. You'll recognize a few from earlier episodes, but this time the month's biggest finds are all in one place. Let's get into it. Clean Your TV helps you clean up an Android or Google TV and replace its crowded home screen with a simple app dock. It connects over Wi-Fi, backs up your settings, and can disable pre-installed apps and add services you don't want.
00:27 You can use a coding assistant to guide the setup or do it yourself. The project also includes a restore tool if you change your mind. Deepsee has open-sourced part of the system it uses to run its AI models at scale. Deep infer handles things like batching requests, distributing work across accelerators, and generating structured outputs such as JSON.
00:48 The release runs DeepSeek V4.1 Flash offline on supported Nvidia and Huawei hardware. You'll still need to build extra components to turn it into a complete online service. Anatomy is an agent skill that draws interactive isometric illustrations of machines and complex systems with no npm dependencies at all. You give it a concept and the agent gathers real numbers, invents a physical object that shows how it works, then builds it with a drawing kit.
01:19 The result is a shaded SVG with live readouts and smooth motion, plus optional WebGL effects for fire, water, and light. Harness Squared from Google Research improves everything around an AI model from prompts and tools to memory and skills while leaving the model itself untouched. For each task, it proposes changes to the harness, tests them, and keeps whatever behaves best with no answer key needed.
01:45 Lessons from those runs are saved so later tasks start from a better harness. OpenAI's math repository is a public dump of research manuscripts written by an unnamed internal model used to test it on open problems in mathematics. The model was given about 4,000 problems and averaged around 3 hours of thinking on each result it kept. That produced 721 manuscripts across 373 problem families, all in PDF and latex.
02:16 Wordcraft is an open- source word processor written in Rust, a clean room rebuild of the classic desktop word processor. It opens and saves dox files, reads old doc files, and exports to PDF. It handles styles, tables, track changes, comments, and mail merge with spelling and grammar checks that work offline. The interface draws natively on the GPU, and the same app can even run in your browser through web assembly.
02:44 GPU unlock is a Linux kernel module that lifts a cap Nvidia places on the tensor cores of its GeForce cards. On these cards, FP16, BF-16, and TF-32 matrix math runs at half the speed the hardware can do. The module loads at boot before Nvidia's own driver and removes that limit without a patched driver or a VIOS flash. It's stable on RTX30 cards and experimental on the 20 and 40 series.
03:11 Replica skill is a set of 11 clawed skills that study an existing app and rebuild its features as a new product of your own. They run in order from recon and architecture to design, build, backend, testing, and deploy. Each picking up where the last one left off. It's clean room, copying features and flows, but never code logos or text. And it scores how close your version's features get to the original.
03:37 Digup is a Mac OS menu bar app that finds files on your Mac by describing what's in them. Built on Google Deep Minds embedding Gemma 2 running locally through llama.cpp. Search in plain words in more than 100 languages and it pulls up the matching photo document or code, even the exact page of a PDF. It can jump straight to the right moment in a video or podcast.
04:02 The everything project from the team behind Pewtor is a single repo that sets out to rebuild the world's common software from scratch as open source. It holds about 40 apps in one place. Office tools like a spreadsheet, documents, and slides. Creative tools for photos, video, vector art, audio, and 3D, plus CAD, circuit design, a CRM, accounting, and invoicing.
04:23 There are even games like chess and sudoku. Free My TV replaces the ad-filled home screen on Google TV and Android TV with a clean one built as a plain Android app with no outside libraries. Your chosen streaming apps sit on a glass shelf. Every hour, it shows three trending picks, but only from apps you already have split into movies and shows. You also get the hourly weather, sunrise and sunset, and a screen saver with a big clock.
04:54 UB mode is a coding agent skill built around Uncle Bob's software engineering principles. It tells your agent to think through the problem, write tests, fix root causes, and refactor as it goes. It also works with companion tools that check code complexity, find duplication, and run mutation tests to see whether your tests catch broken behavior. The goal is to give your AI coding sessions a bit more discipline.
05:21 Horizon is an unofficial rewrite of Minecraft Java Edition in Rust, rendering with WGPU and Vulcan. It contains no Minecraft code or assets. Its launcher builds everything from your own copy of the game. It reads the same worlds, generates the same terrain, and runs the same mobs and redstone checked against the original by differential test harnesses.
05:44 DSBot is a community plugin for DeepSeek Harness that lets you build a team of longived AI bots that remember things. A main bot called the Chief runs things and you can drop several bots into a group chat with one acting as admin. Each bot gets its own memory and its own choice of model. They hand work off to each other and share one library of MCP servers and skills.
06:08 ESP32 iPhone agent lets an AI model use a real iPhone with a board that costs about $7 as its hands. The phone mirrors its screen to a computer over AirPlay, which is how the model sees it. The ESP32S3 board plugs into the phone as an ordinary USB mouse and keyboard, and assistive touch turns every click into a tap. One model plans the steps toward your goal, and another picks each next move from the latest screenshot.
06:37 Pella is a tiny library from the Obsidian team that turns mermaid diagram text into SVG with no dependencies at all. It supports every diagram type in Mermaid 12.1, but draws them with its own layout and visual style. Rendering happens instantly and synchronously, so there's nothing to wait for. Colors and fonts come from CSS variables, so a theme change shows up without redrawing the diagram.
07:04 Zeroach toolkit checks your devices for traces of known spyware, including Pegasus and Predator. It can scan computers, analyze data collected from Android phones, and check iPhone backups. The scan runs offline and produces a report showing what it found and what it couldn't check. A clean result doesn't prove a device is safe, but it can help you decide what needs a closer look.
07:29 Openilm is an open source tool that lets coding agents like Claude Code and Codex make videos as web pages. Every shot is a page with one function that draws the picture at any moment in time. And an HTML timeline lines the shots up into tracks. The same moment always draws the same frame, so renders come out identical every time. A local editor called Studio lets you tweak things by hand or just ask the agent.
07:56 Leanria is a tiny agent skill, just one 8 kilobyte file for reverse engineering apps and binaries built as a lighter alternative to the REA tool set. It doesn't add an MCP server at all. Instead, it tells your agent which ordinary command line tool fits each kind of target from Electron app archives to Mac OS binaries, Go programs, Android apps, and web assembly.
08:20 Day trader is a set of 11 clawed skills for practicing day trading on US stocks written in plain Python using free public data. The skills hand off to each other. One scans charts, another watches the news, then others time your entry and exit, size the position, and check it against your risk limits. Paper trading is on by default, and nothing places an order until you approve that exact trade.
08:46 Earshot gives coding agents the ability to understand sound running on the CPU as a command line tool, a Python API, or an MCP server. It describes what a recording sounds like in plain words, scores how well a sound matches a description, and finds the best moments in a long clip. It can check sound against video cuts, tell a real recording from a synthesized one, and even write code that recreates a sound effect.
09:13 CC BAGA is a desktop toolkit for managing multiple local clawed workspaces. You can archive one identity, switch to another, and choose which settings and conversation history stay with each one. It also includes a separate browser profile and tools to check whether your browser's language and time zone match your network location. It works on Windows, Linux, and Apple Silicon Max.
09:38 Ali brings AirDrop to the Linux so you can send photos and files directly between your computer and an iPhone. Your Linux machine can show up in the iPhone's share menu and it can send files back too. It works with a range of Wi-Fi cards, though compatibility varies. On some cards, your regular Wi-Fi connection pauses during a transfer, so check the hardware list first.
10:02 Hairline gives your website a collection of interactive line illustrations that respond to your cursor. Hover over a keyboard and the keys sync. Move across a phone and its layers pull apart. There are more than 30 figures and you can adjust their colors and how strongly they move. It works with React or plain web projects and there's a coding agent skill for creating your own.
10:24 Tendero is a Mac OS app that hangs your screenshots on a little clos line at the top of the screen instead of piling them on your desktop built in Swift. Rest your pointer in the menu bar and the line slides down with your latest shots. Click one to copy it. Hold to mark it up or double click to open it in preview. Drag a screenshot into any app to drop in a copy or into a folder to take it off the line.
10:54 OXR turns messy PDFs and document images into data your AI tools can actually work with. It picks out text, tables, formulas, and even chemical structures, then exports the results as markdown or JSON with their positions on the page. You can try it through an online demo or run it yourself on a supported NVIDIA GPU. You can also swap in your own layout model for different kinds of documents.
11:20 Netmerge combines several internet connections at once, like your Wi-Fi, Ethernet, and a phone hotspot, so your apps use all of them together. It's a single go program that creates a virtual network adapter with no account, no server, and no VPN. Each new connection goes out over one of your links with faster links getting more, and a dead link stops getting traffic until it recovers.
11:45 TSRS is a Rust port of the TypeScript 7 type checker from the OX C project. Built to be a drop-in replacement for the official compiler and language server. It aims to match TypeScript's results exactly while using far less memory. Install it with PNPM and run it just like TSC or hook it into your editor as a language server. There's also a web assembly build that runs in Node and in the browser.
12:11 Git AI is an open-source version control system designed for lots of coding agents working on the same project. Instead of each agent downloading the whole repository, it can fetch just the files it needs. It batches changes headed for the same branch, keeps a record of which agent made each update, and can use a resolver to propose fixes for merge conflicts.
12:31 It's based on a design anthropic described called Scion. Muse Gadgets lets you build your own devices that connect to the Muse app. Grab an ESP32 board or a Raspberry Pi. Add a screen, buttons, or sensors, and use the SDK to make them work together. You could turn a spare Pi into a gadget for controlling your home assistant setup. The code and firmware are open source, though you'll need an SDK token to pair a device.
13:02 3D model anything from West Boss turns a scan of a room into an editable 3JS scene with an AI agent doing most of the modeling. You capture the room with an iPhone app that records LAR depth and photos. Then a skill walks the agent through measuring the space, building the model, and comparing it against your real photos from the same camera angles over and over until it matches.
13:27 Tstack is Theo's own set of agent skills for Claude code and codeex cleaned up so anyone can use the ones he relies on every day. Each skill fires from a simple phrase, say full send, and an agent proves a change works, gets it reviewed, files the pull request, watches CI and the review bots, then merges. Say council, and you get a second opinion on your change from models built by other companies.
13:53 Manomm GX lets you create animated math explainers and 3D videos using Python. It follows Manim's API, but uses a Rust rendering engine that runs on your GPU. You can combine equations and graphs with 3D objects and moving cameras, then export the result as a video. It also works in the browser and a coding agent can help you write and check the scenes.
14:17 Hatchzip helps you unpack large archives when your drive is nearly full. It extracts the files while shrinking the original archive, freeing up space as it goes. It supports formats like zip, 7z, raar, and tar, and it checks whether the extraction will fit before starting. The archive is used up during this process, but you can choose to keep it if you have enough space.
14:41 Brutal is a C# framework for developers who want to build a game or even their own game engine. It gives you building blocks for graphics, physics, audio, and networking. while leaving the bigger design decisions to you. It's already the foundation beneath Kitten Space Agency and Rim World's Way engine. There's no built-in editor or renderer, so expect to do some hands-on work.