GitHub Trending Weekly #52: shaders, RemoveMacAI, SCSKiller, bloodborne_pc, gdp-ts, open-instinct Transcript

Agent Memory Repo is an open standard and agent skill for carrying context between coding-agent sessions in a separate Git repository. Agents load a short MEMORY.md entry point, follow links to Markdown notes, projects, queries, and scripts, then update and commit memories as they learn. The repository can support personal memory, shared team knowledge, and agent swarms; Git provides version history and flags conflicting edits when multiple agents update the same files. The skill can be installed for Devin CLI or through the skills installer for Claude Code, Cursor, and other agents, and requires Node.js and Git for the latter. A separate persistent local or remote repository is needed for memory across machines or cloud sessions. The specification was developed by Cognition, is open to ecosystem contributions, and is licensed under MIT.

Amane is a Rust library and CLI for building Wayland desktop shells such as bars, panels, and launchers in plain Rust. Shell code lives in a configurable Rust source file, and `amane dev` builds, starts, and rebuilds the shell on saves while keeping the previous shell running if a build fails; separate commands compile, run, and clean the build output. It provides layout, widgets, input, animation, and built-in services for audio, network, Bluetooth, notifications, and workspaces. Amane requires a Wayland compositor supporting wlr-layer-shell, such as niri, Hyprland, or Sway, and is described by its repository as early and experimental.

Animate is a Claude Code skill and plugin for creating short animated videos procedurally in code. It builds a self-contained canvas animation and renders it to MP4 with a headless browser and ffmpeg, using an intake flow followed by story, visual-style, and storyboard approval checks before rendering. The skill includes seven built-in styles—cut paper, crosshatch, risograph, sketchbook, math, pixel, and isometric—and can create an additional style from reference media by measuring its palette, line weight, texture, motion, and cut rhythm. It supports beat-mapped or code-composed music, narrated voice-over timing, real product screenshots, shape-morph transitions, and 9:16, 1:1, 4:5, and 16:9 outputs from one piece. The repository provides command-line tools for building, previewing, storyboarding, exporting, reviewing, comparing references, and checking rendered frames; it is distributed under the MIT license.

ascii is a TypeScript library for adding animated ASCII art to web pages. It provides pieces including scenes, interface elements, charts, text effects, language logos, distribution logos, shapes, physics simulations, nature scenes, and generative patterns, rendered as text in a <pre> element or as colored graphics on a canvas. It supports React, Next.js, Astro, plain HTML through the <ascii-art> custom element, and direct TypeScript usage with its mount API. Pieces can be bundled as modules or loaded by name; options control frame rate and piece-specific settings, while labels support screen readers and reduced-motion users see the first frame by default. The project also provides animated SVG versions of selected logos, companies, and distributions for use in GitHub READMEs. It is written by bas3line and distributed under the MIT license.

Athanor is an MIT-licensed alchemical desktop layer for Arch Linux and Hyprland. It configures seven dungeon-themed workspaces, a roguelike status bar that displays the current workspace as a dungeon level, free disk space as gold, battery as HP, and minutes since boot, plus crash notifications with tombstones. Its bar calculates the planetary hour from the machine's sunrise and sunset and associates each hour with a Golden Dawn tarot trump. It provides four color schemes based on stages of the alchemical Magnum Opus, with an optional setting that changes schemes as the sun moves through the day. It also supplies themes for Helix, Micro, Neovim, and Lualine, and uses two-tone Floyd–Steinberg-dithered Doré engravings as workspace wallpapers. The installer places files in ~/.config/athanor and can install the shell command, prompt, fonts, or an optional Hyprland plugin that displays window titles in borders. Commands include athanor transmute for changing color schemes, athanor draw 3 for a three-card spread, athanor tomb for a crash tombstone, and athanor doctor for installation checks.

bbport is an experimental native Linux port of Bloodborne for PlayStation 4 version 1.09 (CUSA03173) on x86-64 PCs. It converts the game's executable offline into a flat memory image, runs its x86-64 code directly without CPU emulation, and supplies a small runtime implementing the PS4 functions the game uses, including memory, threads, files, audio, gamepad input, saves, and application content. Its graphics layer translates the PS4 command stream and GCN shaders to Vulkan using a renderer derived from shadPS4, with a two-stage GPU command pipeline, game-specific motion-vector generation, temporal upscaling through FSR 3.1, FSR 4, or FSR 4.1.1, unlocked frame-rate patches, and an in-game settings menu. A GTK4 launcher and Steam Deck AppImage provide configuration, controller support, mod and patch selection, and resolution and upscaler controls. The project requires Linux, a Vulkan 1.3 GPU, and the user's own decrypted Bloodborne version 1.09 dump; it includes no game files and is not affiliated with Sony, FromSoftware, or AMD. The new memory and translation mode is experimental and currently requires an AMD GPU. bbport is licensed under GPL-2.0-or-later.

Bops is a Mac app from Nick Vasilescu and Orgo that provides a team of AI bots for business operations such as inboxes, pipelines, invoices and reports. A main bot delegates tasks to specialist bots, which work on Orgo cloud computers with live screens that users can watch or take over. Bots retain workspace context, communicate through the app, Slack, Telegram and Discord, and can use connected services including Gmail, Calendar and about 1,400 other apps. Reading runs immediately, while sending, creating and paying require approval unless the bot is configured otherwise; logins are stored in the Mac Keychain. The open-source repository includes the Electron and Next.js Mac app, the bot server, cloud-computer templates, hosted Bops Cloud components and a relay for self-hosting. Hosted Bops uses OpenAI for chat, tasks and voice and can provide bots with email addresses and phone numbers on eligible plans. Self-hosting requires an Apple-silicon Mac, Node.js, an Orgo account and OpenAI keys, with optional provider integrations. The code is released under the Functional Source License 1.1, with each version becoming available under Apache License 2.0 two years after release.

C64 Keyboard is a font that reconstructs the printed lettering on classic Commodore 64 keycaps from photographs of a Central European machine. It includes uppercase letters, numbers, punctuation, arrows, £, π, selected accented letters, function-key labels, and all 63 PETSCII graphic front legends; lowercase input displays as uppercase because the font represents keycap legends rather than the Commodore 64 screen font. The PETSCII symbols are reconstructed outlines with key identities and assignments based on the Ultimate Commodore 64 Reference, while frame thickness, curves, and print proportions are inferred from the photographs rather than traced exactly. The project provides TTF and OTF desktop fonts, a webfont, editable outlines, build scripts, documentation, and a browser-based specimen and type tester. Its repository contributions, including the font files, outlines, scripts, preview, mappings, and documentation, are dedicated to the public domain under CC0 1.0; the reference photographs and rights in historical keycap designs or trademarks are not included or granted.

Charakuru is a local character-modeling kit that turns a single full-body illustration into a VRM avatar with assistance from an AI agent. It generates eight reference images, creates and processes 3D candidates through Tripo, and uses Blender for part separation, weight transfer, rigging, spring physics, and VRM export. The workflow pauses at checkpoints so the user reviews images, selects candidates, adjusts proportions, and checks motion while the agent handles file management and browser or Blender operations. The kit includes a local review app, seven agent Skills, scripts, documentation, and a base Blender body for weight transfer. It runs on a user's computer and uses external services through the user's own accounts, including Tripo, an agent with image-generation and browser-control capabilities, Blender, the VRM Add-on for Blender, and Blender MCP. The project is distributed through release ZIP files rather than GitHub's automatic source archive and is still in development. Current limitations include neutral faces with fixed gaze, no expressions or eye movement, and physics checks that occur in the exported VRM rather than the in-app walk preview. Its terms permit private, commercial, and non-commercial use subject to redistribution restrictions and additional licensing requirements for businesses meeting the stated revenue or funding threshold.

Claude Code mods is a collection of independently installable plugins for Claude Code, built on hooks that add live status lines, guards, panes, commands and games to the terminal. The mods include a context bar showing token categories and compaction, usage and cache meters, sub-agent and tool-call tracking, Markdown preview with before-and-after views, review monitoring, a map of files touched by agents, and a guard that pauses destructive commands such as recursive removal, force pushes and migrations. Other mods integrate with services or local applications including Spotify, OpenAI, Chrome and external connectors; the repository documents installation through a Claude Code marketplace and individual mod names, and requires Claude Code 2.1.287 or newer, with some mods having additional requirements.

convt is an open-source local file-conversion desktop app and command-line tool from opencoredev. It keeps images, video, audio, PDFs, and documents on the user's machine, registering available native engines such as FFmpeg, LibreOffice, PDFium, image, and resvg, then selecting a conversion route. When no direct conversion is available, it can chain up to three steps. The desktop app provides right-click menus on macOS and Linux, while the CLI supports batch conversion, recursive processing, page ranges, output directories, format discovery, and engine inspection. A document pack enables Word, Excel, and PowerPoint conversion; convt also offers a paid cloud API and TypeScript SDK. The source is licensed under GNU AGPL v3, while signed desktop builds, convt Cloud, and the API are paid.

Durable Actors is an open-source, vendor-neutral runtime for stateful serverless functions used in real-time applications such as chat systems, collaborative documents, and agent swarms. An actor is a class with durable state that survives interruptions, errors, and restarts, together with serialized execution so concurrent callers can update it safely. The runtime loads actors on demand and persists fields marked for persistence; the SDK can generate type-safe Python and TypeScript clients, and clients can connect through WebSockets. Projects can run locally and later self-host the runtime on Google Cloud Platform. The repository describes built-in observability and uses the MIT license.

FreeVideo is an open-source local inference engine and ComfyUI plugin for generating videos with the MiniMax H3 model on consumer hardware. Built on OpenVDN's eight-step VDN-H3 model and Video DeltaNet's hybrid attention, it adapts weight placement, compute precision, attention kernels and acceleration to available hardware. Weight streaming, asynchronous prefetching and chunked computation reduce peak memory use, with the project reporting operation on systems with as little as 8 GB of VRAM and 16 GB of RAM. It accepts text prompts, first and last frames, and image, video and audio references, and supports MiniMax H3 LoRAs, two-pass sampling, batch generation, four quality levels and preview-first generation. Windows and Apple silicon launchers configure ComfyUI, the runtime and models, while Linux supports setup through a shell script and command-line generation. The code is licensed under Apache License 2.0; the model weights use the MiniMax H3 Community License.

gdp-ts is a TypeScript library, linter, and coding-agent skill that applies Ghosts of Departed Proofs to authorization and entitlement checks. It makes sensitive functions require a compile-time proof that a specific user can act on a specific resource, rather than accepting an unconnected boolean permission check. The library's `name()` function gives runtime values distinct compile-time identities, trusted proof modules verify relationships between those named values, and protected functions demand proofs tied to their exact arguments; using the wrong proof, a proof for another value, a raw identifier, or no proof produces a type error. Proofs are represented by small runtime objects while most of the safety comes from TypeScript's type checker, with negligible runtime overhead and no added dependencies for the core package. Its ESLint and Oxlint presets detect practices the type checker cannot, including forged proofs and proof minting outside designated modules. The companion skill installs through `npx skills add rauchg/gdp-ts` and provides a workflow, implementation recipe, linting guidance, examples, and limitations for applying the pattern. The core package requires TypeScript 5.4 or newer and is licensed under MIT.

Genex Desktop is a desktop application for developing games with AI on macOS and Linux. It provides a multi-agent workspace in which a main agent, worker agents, and reviewers can use Claude Code, ChatGPT, or local models; projects can be exported or published to the web. The app supports local 3D asset creation through a Blender plugin and can connect to services such as Meshy, Tripo, and ElevenLabs through its tools router. Its plugin system lets developers add asset generators, engine bridges, and game services, with plugins installable from GitHub or the Genex Marketplace. The project is distributed under the MIT license; Windows, Unity, Unreal, and native C++ support are listed as upcoming.

An open-source long-document search tool that finds pages answering a question by treating page selection as a multiple-choice decision for Jev, without a vector database or embeddings. For shorter documents, it presents each page as an option in one Choice; for longer documents, PageIndex builds a hierarchical tree of document sections and pages, and Jev selects among a small number of nodes at each level. The repository also demonstrates beam search over multiple candidate paths and Noul-based yes/no checking of candidate pages. It accepts PDF documents and questions, uses TypeSafe and PageIndex APIs, and is released under the Apache 2.0 license.

Kandinsky 6.0 Video is a family of diffusion models from Kandinsky Lab for synchronized text-to-audio-video and image-to-audio-video generation. Its Lite and Pro models produce five-second video clips with synchronized 44 kHz audio, including lip synchronization. A separate super-resolution model can raise the output to Full HD. The project supports NVIDIA GPUs and provides integrations for Diffusers, ComfyUI, vLLM-Omni, SGLang, and FastVideo, with command-line and API-based generation workflows.

Leviathan is an open-source command-line tool and optional MCP server for giving coding agents searchable memory over large datasets. Its single static binary indexes JSONL, JSON, CSV/TSV, SQLite, or database-CLI exports into one local SQLite file with an FTS5 index. Searches resolve groups and filters, rank records with BM25 and configured boosts, and return capped result cards with citations instead of exposing raw logs; agents can also resolve groups, retrieve records, inspect the index, and maintain it through upsert and delete operations. It has no runtime dependencies, does not hold database credentials, supports shell and MCP integration, and is licensed under Apache-2.0.

mintocode is an AI coding-agent context tool that compiles a repository into plain-text cards describing files, functions, modules, dependencies, flows, aliases, and likely questions. For each question, it selects relevant paragraphs from those notes, follows a symbol-level link graph to referenced definitions, callers, and tests, and combines the selected notes with current source spans into a compact context pack rather than repeatedly sending whole files. The compiled data is stored in .micode/ and can be updated incrementally; citations are checked against a parsed symbol table and unresolved citations are flagged. The mic command supports one-call questions and pack generation, while integrations provide hooks, MCP tools, commands, and read guards for Claude Code, Codex, ZCode, and OpenCode. It is distributed as a Python package requiring Python 3.10 or later with no dependencies, uses existing Claude, Anthropic, OpenAI-compatible, or Codex model access, and is licensed under MIT.

Mortiflix is a local motion-design studio for producing AI-assisted videos with Claude. It guides a project through briefs, scripts, style frames, transition boards, animatics, builds, and final renders, pausing at approval gates between stages. Reviewers can pin notes to video frames, moments, or script paragraphs; subsequent sessions must address each note before continuing, while approved checks and preferences can be carried across projects. It runs through Claude Code with the user's login or through an Anthropic API key, and provides both a terminal CLI and a local web studio. Built-in pipelines cover explainers, vertical social shorts, and logo animations; renders use Remotion and are checked with ffmpeg. The repository also documents narration through ElevenLabs, local Qwen3-TTS via ComfyUI, user-recorded voice, Strudel-based music generation, asset acquisition through a browser harness, and Blender-based 3D tooling. Users can define additional pipelines as folders containing configuration, craft instructions, and skills. Mortiflix is distributed under AGPL-3.0.

OpenChart is an open-source desktop trading-chart application from longsurf-ai that brings AI agents into market analysis, indicator creation, alerts, and research. Users can connect their own Claude, Codex, or Gemini accounts; describe conditions to watch, and have an agent generate Tea monitors that evaluate incoming market data and investigate when conditions fire. Users can also write custom indicators and conditions in Tea, a language built by the same team for market analysis, and use charts, indicators, Volume Profile, price alerts, and indicator alerts without software-plan quotas. Selecting a chart range lets an agent research the corresponding move and pin annotations with linked sources to the chart, while drawn lines or zones can trigger agent research when price crosses them. The desktop application is free, with AI usage governed by the user's agent-provider plan and optional cloud market data priced separately. OpenChart uses the OpenChart License, based on Apache 2.0 with additional conditions.

Open Instinct is an open-source personal AI agent that users text through iMessage, SMS, email, or webhooks. It runs its own agent process in a microVM with a Linux desktop, uses the Pi agent loop and Composio to operate connected applications, and can create files such as PDFs in its workspace before sending them back as attachments. A policy guard evaluates every tool call. Its network model defines six trust tiers for contacts and their agents, with grants for narrowly scoped exceptions; permissions, invitations, agent-to-agent tools, approvals, scheduling, memory, and audit logging are handled by the repository's packages. Payments use a Stripe Link wallet to request a single-use card for an exact total after owner approval, with spending limits and confirmation rules defined in a policy file. The repository provides a TypeScript CLI, local development and Maritime deployment paths, a multi-user gateway, and documentation for configuration, permissions, protocols, security, and self-hosting. It is distributed under the MIT license; the README describes live iMessage and Maritime deployment as beta and requires users to provide their own service keys.

photoc is a command-line toolkit for photographers that reviews, inspects, organizes, and prepares JPEG photo collections from a terminal. It can interactively mark photos for keeping or rejecting, read EXIF metadata, summarize shoots, group images into sessions, filter by camera and exposure metadata, find blurry or byte-identical files, check JPEG integrity, generate contact sheets, rename and sort files, create compressed copies, and scrub GPS or other supported metadata before sharing. It reads metadata from Sony ARW files but does not develop RAW images; other RAW formats and HEIC are unsupported. Read-only commands leave files unchanged, while rename and sort require --apply and metadata scrubbing requires --in-place; compression, contact sheets, and scrub operations otherwise create new copies. The tool supports JSON output for several inspection commands, runs on macOS and Linux, is distributed through a Homebrew tap or verified shell installer, and is licensed under the MIT License.

Pi Pocket is a self-hosted, mobile-first web application for running and supervising Pi coding agents on a user's own machine or phone. It stores model calls, tool calls, subagent activity, sessions, notes, schedules, and other state in a durable database so work can continue after server restarts; the web client receives committed session updates through server-sent events. The app supports live multiplayer sessions with presence, side chat, shared notes, mentions, reactions, pins, invitations, roles, and turn-taking. Users can steer or queue tasks, inspect tool output and diffs, fork or retry sessions, use git branches and worktrees, approve plan-mode changes, schedule messages, run goals until a check passes, and open a Chromium-based browser panel. It also provides artifacts, subagents, push notifications, extensions, codemode, and an optional Lancet Guard that requires approval for selected risky calls. Pi Pocket runs on Linux, macOS, Windows, or Android through Termux, requires Node.js 22.19 or newer, and can expose the app locally, over a LAN, through a Cloudflare quick tunnel, or through Tailscale. It is distributed under the MIT license. The project warns that anyone with steering access can make Pi run commands with the permissions of the host machine, and that Pi Durable is experimental.

P.T. PC Port is an unofficial native Windows and Linux port of Kojima Productions' 2014 PlayStation 4 teaser P.T. Its game logic is rebuilt in C++, while the renderer uses Vulkan; the port reads the levels, models, textures, sounds, scripts, and cutscenes at runtime from the user's own PS4 game files rather than distributing game data. It plays the complete teaser, including the voice section, and adds PC features such as configurable resolutions, upscaling, ray-traced graphics options, photo and free-camera modes, a loop browser, a museum of discovered content, speedrun timing, mod support, and experimental OpenXR VR. It requires a compatible game dump or fake PKG, a Vulkan 1.3-capable system, and Windows 10/11 or a recent 64-bit Linux distribution. The port's own code is licensed under the MIT license; it is not affiliated with Konami or Kojima Productions.

RemoveMacAI is a macOS debloating and privacy tool from Om Lahore, distributed as a native app and command-line program for supported Apple Silicon Macs running macOS 27. It can disable selected Apple Intelligence features, remove their models and block future downloads, turn off analytics, personalized ads and other settings, silence pop-ups and background items, and clean up selected macOS and developer files. Changes are shown for review before application and can be undone through a shared history; the CLI can also export configuration profiles for MDM deployment. Locked settings are managed through a configuration profile, while other preferences are recorded in a journal before being changed, and storage cleanup moves files to the Trash except for certain simulator and Time Machine operations. The project states that it keeps System Integrity Protection enabled, modifies nothing under /System, makes no network requests and collects no data. It is released under the MIT license.

SCSKiller is a free, open-source Windows tool that pre-compiles a game's shaders into the NVIDIA or AMD GPU driver's cache before play, reducing stutter caused by in-game shader compilation. It reads the shaders shipped with a game, determines the pipelines to build, and compiles them in a separate process while the game is closed; an optional recorder can capture pipelines encountered during play when the files do not provide them. It supports DirectX 12 games and DirectX 11 games on NVIDIA, detects games from several PC launchers or from user-added locations, and can recompile caches after driver updates. The tool does not modify game files during normal operation and leaves anti-cheat games alone apart from reading their files, with documented opt-in exceptions for offline sessions of Elden Ring and Armored Core VI. It is distributed for 64-bit Windows 10 and 11 under GPL-3.0-or-later, with an additional permission covering certain bundled components.

Shaders is an open-source WebGPU component library and design platform from Shader Effects, Inc. It provides declarative effects for React, Vue, Svelte, Solid, JavaScript, and Framer, including gradients, noise, glass, metal, lighting, distortions, transitions, blurs, and cursor effects. Components render as GPU-blended layers that can be nested, masked, and composed without writing shader code. Its visual editor at shaders.com lets users arrange components on an infinite canvas, adjust their properties, connect them to cursor or timeline inputs, and export the resulting component tree. The accompanying CLI connects a project to a Shaders account, detects the framework, installs selected effects as component files, and can search, preview, and update them. Custom components can be defined with JavaScript objects and WGSL pixel code. The engine, components, framework bindings, and npm package are MIT-licensed; the hosted editor, presets, and other platform features have separate terms.

Shipyard is an open-source desktop app and toolset for running coding agents on remote development machines such as VPSs, cloud VMs, and desktops while controlling them from a laptop. It gives each repository worktree terminals, browser tabs, dev servers, services, hooks, and automations; agents run on the remote machines, so closing the app or losing the laptop connection does not stop them. Shipyard supports agents including Claude Code and Codex, with a kanban for agents needing attention, a review inbox, and orchestration actions such as prompting, waiting, checking, looping, and handing work off. Projects can be configured through .berth/config.json, shareable kits, or machine configuration, while automations can run from events, schedules, or GitHub activity. Its components include the berthd daemon on each development machine, the berth laptop CLI and background agent, a Tauri and React desktop app, and React plugins. The repository provides installation scripts and prebuilt desktop and daemon releases, SSH-based machine pairing, private development-server URLs, and integrations that report Claude Code, Codex, and Cursor state. It is distributed under the MIT license.

showtime is an open-source local video studio for coding agents. It turns a one-sentence video request into motion graphics, voice-over, music, sound effects, captions, footage edits, platform exports and self-contained HTML videos, running on the user's machine without its own cloud service, API keys or media uploads. Motion graphics are authored in HTML, CSS and canvas and rendered frame-by-frame in headless Chrome; mathematical scenes can use Manim, while footage is transcribed and edited with FFmpeg. The agent shows stills or a fast draft before the full render, then runs checks for issues such as loudness, black or frozen frames, captions and platform specifications. Studio mode provides a local board for selecting concepts, visual styles, music and storyboards, and optional specialist sub-agents can handle research, scene building and critique. The project is distributed as an Agent Skill, command-line tool and MCP server for hosts including Claude Code, Codex, Cursor, Devin and OpenCode. It downloads models and tools into a local ~/.showtime directory, records job receipts and attribution credits, and creates separate output folders for each job. The repository's own code is MIT-licensed; downloaded third-party tools, models and media retain their respective licenses.

SLOOP is free, open-source live groovebox firmware for the M-VAVE FM-1, based on Felucca by Leo Kuroshita and Hügelton Instruments. It turns the device into a four-track instrument with three synth tracks and a drum track, ten synthesis engines including six-operator FM with DX7 patch import, live recording, looping, song sections, effects, samples, and a sequencer. Its sequencer supports per-track lengths and divisions, chords, ties, slides, swing, ratchets, parameter locks, micro-timing, fills, and MIDI output. SLOOP supports USB and TRS MIDI, MIDI clock synchronization, class-compliant USB audio recording, and a browser-based editor for sound parameters, sequences, drum kits, samples, projects, backups, and restores. It is distributed under GPL-3.0-only, remains beta software, and can be installed or reverted to the official FM-1 firmware through a browser-based installer.

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.

ts-rust, also distributed as tsc-rs, is an experimental Rust port of Microsoft's native TypeScript 7 compiler. It includes the compiler, type checker, language server, and API, following the upstream Go compiler's command-line behavior and algorithms. It can be installed from npm with `tsc-rs` and used with `npx tsc-rs -p tsconfig.json`; standalone binaries are provided for Linux x64 and macOS arm64, while Windows and Linux arm64 are not yet available. The port also includes Effect language-service diagnostics when enabled through the corresponding TypeScript plugin configuration. The project is an early release with documented compatibility issues involving some monorepos, project-build output, and long editor sessions, and is licensed under MIT.

WERM (Wiring-Encoded Recurrent Modulator) is a computational-neuroscience simulation that models a 302-neuron Caenorhabditis elegans network using the worm's published connectome. It runs as a rate network with a liquid-time-constant-style gate; the model treats the 28 GABA-releasing neurons as inhibitory and the remaining neurons as excitatory, and supports simulated sensory stimuli such as nose and tail touch. A steering layer maps the simulated worm state to language-model sampling settings and a tone prompt. WERM can run in a browser, in Node.js, or connect on the user's computer to Ollama or another local server implementing the OpenAI chat-completions API. The repository reports that tail and nose stimuli produce the expected forward or backward network responses in its tests, while the effect of the language-model steering has not yet been measured. The project requires Node 20 or newer, has no runtime dependencies, includes scripts for rerunning parameter sweeps and experiments, and is licensed under the MIT license. It is an intentionally small model without muscles, body physics, neuropeptides, learning, or comparison against recordings from living worms.