GitHub Trending Today #43: Kakehashi, qm, gander, SimpleEnglish, skill-recorder, humanizer-cli, crm Transcript, AI Summary & Key Points

Anthropic is an AI research and deployment company founded by former OpenAI researchers (including Dario Amodei). It develops large language models such as Claude and focuses on AI safety research and commercial model deployments.

BC7215 AC Matter is an open-source Matter air-conditioner controller built around an ESP32 and the BC7215 infrared controller. It learns commands from an existing air-conditioner remote, then exposes controls such as power, temperature, and fan settings to Matter-compatible platforms including Apple Home, Google Home, and Home Assistant. The controller is designed for old or new air conditioners of different brands, requires no dedicated app or account, does not require access to the air conditioner's internal circuit, and operates offline for the air-conditioner control function. The repository notes that Matter's air-conditioner support is limited and that smartphone platforms may provide delayed feedback or incomplete mappings for split-system features.

Bindwidth is a browser-based calculator for sizing on-premises LLM inference infrastructure and estimating total cost of ownership. It evaluates three constraints for a resident text model—KV-cache memory, combined prefill and decode serving capacity, and the runtime session ceiling—and identifies which constraint binds for a workload. The tool accounts for interactive users and autonomous agents as different load types, then compares owned hardware, eligible sovereign rentals, and enterprise subscriptions with per-token APIs for agents. It distinguishes measured from estimated hardware and model data through confidence levels and profile-specific overrides, flags domain and model-capability limits, and exports a decision record as Markdown or the full scenario as JSON. The repository describes it as a directional sensitivity estimator rather than a procurement quote. It is a frontend-only application with no build step, server, database, account, or telemetry; calculations run in the browser and entered data is not transmitted.

Cata-centavo is a self-hosted MCP server that gives compatible AI agents read-only access to a user's Brazilian Open Finance bank, credit-card, and investment accounts through Pluggy. It runs locally over stdio, reads existing Pluggy connections identified by environment variables, and supports queries about transactions, spending categories, account balances, current card statements, unfamiliar charges, and investments. It has no hosted or multi-user service. The package runs with Node.js 22.13 or newer and can be launched with npx. Its main commands include the MCP server itself, `init` for checking credentials and configured connections, and `doctor` for diagnosing connection status, consent, local cache, and learned categorization data. Provider transaction categories are copied into a persistent local `data.db`; the server also learns merchant-category associations from the user's transactions, and manually corrected or newly assigned categories are retained locally and applied retroactively.

Cloudflare is an internet infrastructure and security platform that provides hosting and delivery infrastructure for applications, agents, websites, and data. Its network combines compute, connectivity, and security at edge locations, with Workers running code close to users and backend data; the company states that its network operates in more than 335 cities and reaches 95% of the world's Internet-connected population within 50 milliseconds. Cloudflare also provides AI crawler controls that let site owners inspect, allow, or block crawler activity, along with pay-per-crawl and a monetization gateway for pages, data sets, APIs, MCP tool calls, files, and search indexes. The described x402 flow uses an HTTP 402 response to request payment, after which an agent pays, retries with proof, and Cloudflare verifies the payment at the edge.

Comp AI CRM is an open-source, self-hostable customer relationship management system for AI agents, developed by Comp AI and hosted by Try Comp AI. Its agent runs as an independent durable deployment on its own schedule and database-backed work queue rather than waiting for browser requests: it selects records to investigate, researches contacts and companies, spends a research budget, schedules follow-ups and rechecks, records observed evidence, and sends weak or ambiguous matches to a human instead of writing them directly to records. It supports durable sessions, contact and company records, an Agent tab for viewing work and answering questions, authored tools for reading CRM history, searching records, identifying contacts, researching people, enriching companies, recording facts and scheduling rechecks, and versioned Markdown skills. The agent uses file-based tools and Markdown skills on Vercel's eve durable-agent framework, while its queue leases due tasks with PostgreSQL row locking so concurrent workers claim disjoint work and expired leases release tasks from failed runs. A restricted shell sandbox provides bash, grep, glob and a workspace without network egress, database credentials or direct database access. Optional sources include mailbox history, company brand data, LinkedIn and Perplexity web research.

Context Ontology Accelerator is an open-source semantic context layer for AWS and AI agents. It combines knowledge graphs, formal ontologies, rule-based systems, and modern AI in a Scan → Model → Serve workflow: it connects data sources, discovers schemas, enriches metadata, and ingests documents; induces and manages ontologies, defines metrics, and builds a unified semantic graph; then serves context through SPARQL federation using a Virtual Knowledge Graph, knowledge-graph traversal, queries, and MCP tools. The repository includes source-ingestion services for databases and documents, ontology induction and reasoning with HermiT and ELK, an Ontop-based Virtual Knowledge Graph, a context manager for query orchestration, and a React frontend. Access is controlled through namespace isolation and role-based authorization, with namespace-scoped roles and platform-level roles. The implementation uses Python and TypeScript, AWS CDK, Smithy-generated API contracts, and an Nx-managed monorepo; the project is licensed under Apache License 2.0.

Cursor is an AI-powered coding agent and integrated development environment designed to accelerate software development by handing off coding tasks to AI. It evolved from an email client into a multimodel development tool, supporting broader developer workflows. The platform also offers MCP-connected capabilities for tasks like searching and editing notes.

eve is an AI agent framework for building durable, structured, multi-agent projects from declarative configuration covering instructions, skills, tools, channels, schedules, permissions, and evaluations. Its agents can be developed within a single folder and run through chat surfaces such as Slack or the eve development TUI. Vercel uses eve as the foundation for a marketing-team template with one lead and five specialists: product marketing, content marketing, social media, SEO, and email. The lead loads shared brand context and user preferences, writes self-contained briefs, routes work one level deep, and returns deliverables produced in Notion, Typefully, or Resend; specialists research and edit their own work against written rubrics rather than spawning further agents. The template uses MCP connections for Notion, Typefully, and Resend, Vercel Blob for shared state and files, Vercel AI Gateway for model access, and Vercel Sandbox for reference files and shell commands. Irreversible actions—including email sends, deletes, scheduled social publishing, and Notion page moves—pause for human approval, while the available Resend tools are restricted to keep account administration out of reach. The project is implemented in strict ESM TypeScript and includes commands for local development, validation, type checking, discovery diagnostics, and deployment.

Gander is a small, open-source, fully offline Android file viewer for PDF, Word, Excel, PowerPoint, photos, video, audio, Markdown, text, and code. It is designed to operate without app permissions, internet access, accounts, ads, tracking, or analytics; files are rendered on the device in an APK of about 5 MB. The app uses offline renderers including pdf.js and docx-preview in a sandboxed WebView, SheetJS for spreadsheets, PPTXjs for PowerPoint files, tiled decoding for deep-zoom images, and Media3 ExoPlayer for video. It provides recent-file thumbnails, folder browsing through one-time system grants, Android share-sheet and “Open with” integration, in-document search, PDF text selection and copying, screen-reader support, and file sharing or location access. The page states that it supports Android 8.0 and later.

Git is an open-source, fast, scalable, distributed revision control system. It stores project changes as versioned commits and provides repository history, diffs, branching, isolated work trees, and rollback capabilities for reviewing or reverting changes. The project provides both high-level version-control operations and access to its internal mechanisms. It is covered primarily by the GNU General Public License, with some components under compatible licenses, and was originally written by Linus Torvalds with contributions from a wider developer community.

GitHub Copilot is an AI-powered software-development tool created by GitHub, owned by Microsoft, in collaboration with OpenAI. It suggests code completions, whole-line and whole-function snippets, documentation, and software changes for review by an experienced engineer in supported editors and IDEs. Its workflow includes inspecting changes line by line, assessing architectural correctness, and coordinating work with other pull requests. It is offered as a commercial subscription integrated with Visual Studio Code, Visual Studio, JetBrains IDEs, and GitHub Codespaces.

Graybox is a local-first long-term memory and personal knowledge-management tool that stores notes as Markdown files with YAML frontmatter. It preserves an immutable inbox of raw captures while an LLM extracts people, projects, tasks, actions, meetings, events, decisions, and other entities into typed, cross-linked pages with backlinks and source citations. Its question-answering function retrieves evidence from captured notes and linked pages and is designed to refuse unsupported answers. Graybox uses keyword search without requiring a vector database, with optional embeddings, and uses LiteLLM to connect to hosted or self-hosted language models. It is distributed as an open-source command-line and terminal user-interface application, with a PyPI package and a GitHub repository.

Grok is a consumer-facing AI assistant and chatbot developed by xAI, providing conversational assistance, personalized responses, image generation, code writing, and real-time answers sourced from the web and X. It is available through web and mobile experiences and integrated with the X platform.

Headroom is a local-first macOS menu-bar app for monitoring AI coding quotas and software delivery status, with optional iPhone, iPad, Apple Watch, and ESP32 desk-display clients. A Python host running on the Mac reads the authentication data and command-line tools already configured for services such as Claude, Codex, Cursor, GitHub, and Vercel, then serves a single local JSON feed; credentials remain on the machine and no Headroom cloud account is required. Its interfaces show provider quota meters, usage rings, burn rate and spend, failed Actions, deployments, monitors, and other attention items, and the iPhone client can approve, deny, or reply to Claude and Codex requests without starting work. The Mac host uses port 8737, requires macOS 14+ and Python 3.9+, and the optional desk display uses a Waveshare ESP32-S3-Touch-AMOLED-1.8 board.

A local command-line reference and checker for common signs of machine-written prose. It provides before-and-after examples, draft scanning, searchable patterns, and an editing guide.

Ink Tide is a cel-shaded arcade boat-racing game for the browser, built with Vite, TypeScript, and Three.js. It uses no external assets: meshes, textures, rider animations, and sounds are generated in code. The game runs on an infinite procedural ocean formed from six summed Gerstner waves displaced in a vertex shader on a camera-recentered radial grid; CPU buoyancy samples the same wave field at six hull points, while GPU water rendering receives shared wave uniforms. Boats race three laps around a closed circuit with checkpoint gates, wrong-way detection, split times, AI drivers, powerslide-charged boosts, camera cycling, and gamepad controls. Rendering uses quantised cel shading, banded specular highlights, Fresnel rims, inverted-hull exterior outlines, and a Sobel pass for interior lines. Audio is synthesized with Web Audio, including engine, water, impact, and race-start sounds. The repository includes a Playwright-based screenshot harness for capturing and verifying game frames.

Kakehashi is a userspace macOS ARM64-to-Linux aarch64 translation layer for running Darwin Mach-O command-line binaries on ARM64 Linux without a virtual machine, JIT compiler, or instruction emulator. It loads Mach-O files, maps a freestanding libSystem environment, and translates BSD system calls at the guest–host boundary while handling guest threads and faults. The CLI provides live execution through `kh run` on Linux aarch64, plus a dry-load mode available on other hosts. It requires a manually assembled macOS guest filesystem, called a bottle, at the fixed path `~/.local/share/kakehashi/bottle/`; the project documents copying core directories from macOS and installing Apple Xcode Command Line Tools into the bottle. The repository states that the project is no longer maintained.

Kimi K3 in C is a portable C99 inference engine for running the 2.78-trillion-parameter Kimi K3 base model on CPUs without BLAS, an inference framework, or a GPU. It reads the 1.56 TB checkpoint from disk, keeps a selected depth of the 93-layer dense trunk in memory, and streams the remainder during generation; its routed experts are packed in 4-bit form, multiplied directly from that representation, and managed with an LRU cache. The implementation includes KDA attention with a non-growing memory, MLA attention with a latent representation, and selection of 16 experts from 896. The repository provides a command-line interface, weightless and full-checkpoint validation gates, and measurement data; it reports operation from 8 GB of RAM upward, with additional memory changing speed rather than the generated output. The base model has no chat template, so prompts produce continuations rather than chat replies. Reference requirements include an AVX2-and-FMA CPU, about 1.7 TB of free storage, and a C compiler; Linux/x86-64 is the reference platform, while macOS/arm64 and Windows/x86-64 are also documented as building and passing the tests.

An open-source marketing-agent team from Vercel Labs built on the eve framework. Users submit launch planning, copywriting, or SEO requests through Slack or the eve terminal TUI; a lead agent loads the shared brand-context document, writes a self-contained brief, routes it to one of five specialists—product marketing, content marketing, social media, SEO, or email—and returns the resulting deliverable. The lead does not write deliverables itself, and specialists work independently without shared conversation history; the product marketer alone maintains the brand-context document, which the other specialists read at the start of a task. Newsletter work passes from the content marketer to the email specialist for inbox adaptation and sending. Outputs are delivered through Notion, Typefully, Resend, or stored audit artifacts. The system pauses irreversible operations—including Resend sends and deletes, Typefully deletes and scheduled publishing, and Notion page moves—for approval in Slack or the terminal; its email specialist is restricted to an allowlist of Resend tools. Deployment provisions Notion, Resend, and Slack connectors and a Vercel Blob store, and requests a Typefully API key. The implementation uses TypeScript, Vercel Connect, MCP connections for Notion, Typefully, and Resend, Vercel AI Gateway for model access, and Vercel Sandbox for reference files and shell commands. Email campaigns require a verified Resend sending domain and at least one segment configured in Resend.

NIGHTRUN is a local LLM runtime written in Rust that boots directly from a USB stick on ordinary x86_64 PCs or from an SD card on a Raspberry Pi 5, without a conventional operating system underneath. Its UEFI application loads a quantized model into RAM, renders a chat terminal through the framebuffer, and runs inference locally and offline without a network stack, host process, or kernel. The installer builds and flashes a bootable image, guides model selection and verified downloads, and supports GGUF inspection and conversion into NIGHTRUN's .nrm container. During boot, the model is streamed into RAM with checksummed sections; storage is then sealed so generation cannot read from disk. The runtime includes custom framebuffer rendering, USB keyboard input, multi-core inference, serial diagnostics, hand-written quantized kernels for x86_64 and Raspberry Pi architectures, and Q8_0, Q4_K, and Q6_K weight formats. It requires Linux, Rust, and approximately 6 GB of free disk space for the model and image.

Notion is a productivity and collaboration platform developed by Notion Labs Inc. It combines notes, documents, relational databases, kanban boards, calendars and wikis into a single workspace for individuals and teams.

OpenAI Codex is an AI coding agent from OpenAI available as a command-line tool (Codex CLI) that helps developers produce software. It can be used alongside Gemini for adversarial audits of software requirements and implementation plans, listed as a supported coding-agent or model option in several projects, and its logs can be joined with task and test evidence. The Codex CLI can also receive and answer requests from the Penako canvas.

OpenDerm is an open-source robotic imaging system for creating high-resolution, reproducible 3D maps of skin and comparing scans of the same site over time to detect early signs of skin cancer. Its three-computer architecture uses one Raspberry Pi with an Octopus Pro/Klipper controller and Pico bridge for the X, Y, and Z axes, a second Raspberry Pi for the CAN-controlled rotation axis, Canon EOS R7 camera, distance sensors, homing, and standoff regulation, and a workstation for 3D reconstruction and longitudinal comparison. The documented workflow covers calibration, scanning, image capture, reconstruction, registration, uncertainty analysis, and comparison. The software requires Python 3.11 or newer and includes motion-control services, camera-capture tools, collision guards, and command-line controls for the robot axes.

Pluggy is a Brazilian open-finance infrastructure provider offering commercial APIs to access bank account and transaction data and to initiate payments and transfers across Brazilian banks. It connects to institutions in Brazil's Open Finance ecosystem and is distributed as a SaaS API used by fintechs and financial applications.

Python is a programming language and the implementation environment maintained in the CPython repository. The videos describe its use for back-end applications, bots, scientific analysis, API scripts, file-operation primitives, database and vector-search result processing, coding-session dashboards, document-converter bindings, and compiled agent skills. CPython can be built on Unix-like systems and Windows, with optional profile-guided and link-time optimization; installable distributions and documentation are provided through python.org.

QM is a multiplayer agent harness for startups that operates through Slack and a web interface. It gives each employee and room an isolated scope with its own memory, files, keychain view, permissions, scheduled jobs, web apps, and durable sandbox, while supporting collaboration in channels, group messages, and projects. Shared skills can be granted by scope, promoted by administrators, or imported from Git repositories; background work can run through crons, watches, and inbound webhooks, and internal apps can be published to selected users. A headless TypeScript core runs directly on Node with Fastify and handles identity, policy, scheduling, persistence, and the agent loop. Deployments can select harnesses and models including Pi, OpenCode, Codex, and Claude Code. PostgreSQL stores sessions, memory, queues, and other durable state, while a fixed tool surface includes an execute tool that runs commands in each scope's isolated, persistent sandbox. Slack is an optional in-process Bolt plugin; the web UI, admin panel, and public portal are optional HTTP API plugins built with Vite and Lit. Each deployment keeps organization-specific configuration, tools, skills, sandbox images, and infrastructure in a deployment directory validated and deployed by the qm CLI. Administrators can set organization-wide configuration, available harnesses and models, and a security posture. Strict mode pauses harness tool calls for human approval, Auto mode screens provenance-labelled external data and tool results with a classifier, and Dangerous mode disables content screening and pauses; a predeclared command policy with approvals and hard denials for operations such as recursive deletion or destructive SQL applies in every posture. Deployments run in the operator's own cloud account and are initialized from the @yc-software/qm package without requiring a source checkout; the project documents its threat model, operator assumptions, and known limitations in SECURITY.md.

QRFerry transfers files between two devices as a live animated QR stream. The sender and receiver run in the browser, and the file is not uploaded to an application server. The protocol optionally compresses the file with Brotli or gzip, then divides the payload into QR frames containing binary headers, RaptorQ source or repair symbols, and CRC-32 checksums. Fountain coding lets the receiver reconstruct the file from enough unique symbols while tolerating blurred, missing, out-of-order, or mid-stream frames. The receiver verifies frame, compressed-payload, RaptorQ-object, and original-file checksums before saving the file. QR rendering, scanning, and RaptorQ encoding and decoding use WebAssembly implementations of fast_qr, ZXing-C++, and an RFC 6330 RaptorQ implementation.

Rabbet is a from-scratch C++ 3D game engine with a sparse-set entity-component system, versioned entity handles, dense component pools, and a system-and-resource runtime. Its OpenGL 4.1 renderer provides a forward path with metallic-roughness PBR, a Phong fallback, lighting and shadow maps, emissive surfaces, optional image-based lighting, post-processing, CPU particles, and heightmap or noise terrain. The Forge editor provides a dockable ImGui workspace around a render-to-texture viewport, asset thumbnails, hierarchy reparenting, click-to-pick, snapping gizmos, and Play/Pause/Step/Stop controls that snapshot and restore scenes. Assets use stable UUIDs and .import sidecars; models, textures, audio, materials, shaders, prefabs, and scenes are supported, with scenes stored as readable JSON. Lua behaviors run through sol2 with live hot-reload and inspector-editable fields, while physics uses Jolt rigidbodies and audio uses miniaudio. The repository includes the engine, editor, examples, and tests, including the Orb Hunt demo game. It uses CMake 3.24 or later and a C++20 compiler, supports macOS and Linux, fetches pinned dependencies during configuration, and is licensed under Apache License 2.0.

Ratchet is a code-auditing tool for examining coding-agent edits. It checks changes for new dependencies, duplicate helper code, thin wrappers, and reimplemented standard-library features, while tracking file, dependency, and line-count budgets in an auditable ledger.

Resend is a cloud-hosted email delivery service for developers that sends transactional and marketing email at scale. Operated by Resend, it accepts email content and recipient data and produces delivered messages plus delivery metrics and deliverability tooling for websites and applications. The product is presented as an API-first service for integrating email sending into developer workflows.

Scopey is a lightweight Rust command-line tool for keeping Claude Code, Codex, Grok, Pi, and OpenCode coding-agent sessions aligned with the user's current intent. It turns prompts into a current scope, evaluates agent tool activity against that scope, and surfaces sessions that need attention when work drifts. It can inject a correction into an active session and send notifications, while ignoring subagent sessions by default. Scopey is pre-1.0 software and is distributed as a Homebrew package or Rust source installation.

SimpleEnglish is an open-source agent skill that makes large language models write technical documentation in ASD-STE100 Simplified Technical English, a controlled language used in aerospace. Its rules enforce short sentences, active voice, explicit instructions, simple tenses, and consistent terminology when rewriting READMEs, error messages, incident reports, runbooks, and release notes. The skill is distributed as a dependency-free folder for tools that support the Agent Skills standard, including Claude Code, Cursor, VS Code Copilot, OpenAI Codex, Gemini CLI, Goose, and OpenCode. It can also be installed as a Claude Code plugin and output style, or used by adding its prompt to a system prompt, AGENTS.md, or .cursorrules file. The repository is licensed under MIT.

Skill Recorder is a Microsoft desktop application that records a work session—including screen activity, application and window switches, browser pages, clipboard previews, and optional spoken narration—and uses the GitHub Copilot CLI to reconstruct the session as an overall intent and ordered steps. Users can review and edit the analysis, then generate either a reusable SKILL.md procedure for an AI agent or an Automation that runs on a schedule or trigger. Generated procedures prefer an agent’s native tools, such as the gh CLI or web_fetch, over replaying interface clicks and can generalize from the recorded example. Recording, frame extraction, local storage, and optional narration transcription occur on the computer; choosing Analyze sends the event timeline, extracted screen images, and narration text to GitHub’s cloud for Copilot processing. Narration is transcribed on-device using a Whisper model. The application is distributed as a source release that downloads a pinned Node.js runtime and builds the release locally without a global installation. macOS is the primary target, with Windows 11 support for x64 and ARM64 and Ubuntu installation instructions. A GitHub account with Copilot access is required.

Sol Advisor is a Codex-only orchestration workflow and plugin for capability-routed software delivery. Its primary Sol / High session receives the goal and constraints, declares a risk-gated route before using task tools, and owns planning, implementation or delegation, verification, and acceptance. The workflow supports solo, delegate, audit, and full routes: solo is the default; delegation assigns bounded or higher-risk implementation to a Luna or Terra auxiliary role while the root session verifies the result; audit and full routes add a fresh, read-only Sol / High review, and any requested fix requires another review. Auxiliary work substitutes for root work rather than duplicating it. The project limits auxiliary work to one by default and requires an explicit exception for the full route. It is installed as a Codex plugin with a companion script that installs and verifies the role files in a fail-closed manner.

Syncular is an offline-first SQL synchronization toolkit that keeps an SQLite database on each client while treating a server-owned commit log as authoritative. Applications read and write local SQLite immediately; writes enter an outbox and are checked and reconciled by the server during synchronization. The browser client runs SQLite on OPFS, while native integrations share a Rust core with native SQLite through bindings for Rust, Swift, Kotlin, Flutter, React Native, and Tauri. Servers can run on Bun, Node, or Cloudflare Workers with SQLite, Postgres, or D1. The project also provides server-side persistent SyncClients, database-less SyncRemoteClients, schema and query code generation, React hooks, column encryption, Yjs CRDT integration, and TypeScript and Rust implementations checked against a shared conformance catalog. Syncular is maintained by Benjamin Kniffler and is pre-1.0; its public APIs and protocol details may change before 1.0. The repository is released under the license stated in its repository files.

An OpenAI repository containing Lean 4 formalizations of ten results presented in its paper on advances in mathematics and theoretical computer science. The formalizations cover high-dimensional sphere packing; binary and spherical codes; non-sofic groups; Connes’s rigidity conjecture; arithmetic circuit complexity; quantum parallel repetition; the closest vector problem; Ehrhart’s volume conjecture; multicolor Ramsey numbers; and compactness and degeneracy conjectures in extremal graph theory. Each result is implemented as a separate Lean module, including `SpherePacking.lean`, `MetricCodes.lean`, `NonSoficGroup.lean`, `ConnesRigidity.lean`, `Permanent.lean`, `QuantumParallelRepetition.lean`, `GapCVP.lean`, `EhrhartVolumeInequality.lean`, `MulticolorTriangleRamsey.lean`, and `CompactnessAndDegeneracy.lean`. The project uses Lean 4.32.0, mathlib, and Lake. With the mathlib cache installed, all formalizations can be built with `lake build All`, or individual modules can be built by name; the repository also provides reasoning walkthroughs and documents independent checking with Comparator.

Termixer is a terminal-based two-deck DJ mixer for Unix, built in Rust with ratatui for live performance with TidalCycles. It controls audio from MPV and SuperCollider and provides deck faders, panning, three-band EQ, low- and high-pass filters, crossfading, cue output, microphone input, and a 4x4 sample-pad grid with sequencing. It can record deck, master, microphone, synthesizer, or pad output to MP3, WAV, FLAC, or OGG, and supports SuperCollider SynthDefs, a built-in sine synthesizer, Vim-style navigation, and JSON import/export for pad and sequence sessions. Installation is available through crates.io, GitHub release archives, or source; MPV and FFmpeg are required for the corresponding playback and recording functions, while SuperCollider is optional for TidalCycles integration.

The Complete Shelf is a single-file Three.js interactive library containing seven clothbound hardcover volumes themed around Codex, Claude Code, Cursor, Antigravity, Figma, Framer, and Xcode. Users navigate a continuous shelf with wheel, keyboard, buttons, or position markers, then move a volume into a responsive detail view with orbit, pan, zoom, hover-to-open, and curved page-turn interactions. Each book is assembled from reusable geometry with separate boards, spine, hinges, endpapers, page block, headbands, bookmark, foil, and contact shadows; procedural textures and book-specific color systems drive the materials and scene. A small state machine coordinates deterministic shelf-to-detail transitions, while accessible HTML controls and status announcements overlay the WebGL scene. The entire experience is contained in index.html, including layout, shaders, geometry, interaction logic, embedded image atlases, and audio; it uses Three.js and has no backend, bundler, analytics layer, or browser-side API calls.

Trace File Lineage is a local-first Python and command-line tool for tracing the origins of files, recording task outputs, and helping place new outputs in an existing workspace layout. For existing files, it reads source code, documents, image metadata, and Git history to rank possible scripts, notebooks, inputs, and runs, attaching evidence and marking results as uncertain when proof is unavailable. For future runs, it wraps a command and records verified provenance from the files changed within the task boundary. It can produce a complete manifest with a nested changed-file structure, show ranked lineage candidates, identify files affected by an input, report stale files, and suggest an output path based on workspace conventions. It is distributed as a Python package with the `trace-file-lineage` installation name and includes a demo command that writes to a local demo directory.

Typefully is a SaaS social media composition, scheduling, and analytics platform developed by Typefully that publishes posts to X (Twitter), LinkedIn, Threads, Bluesky and other networks. It accepts text, images, links and scheduled metadata as inputs, provides drafting, queueing, scheduling, publishing, basic analytics and automations, and outputs published posts and performance reports. The service is offered commercially with free and paid tiers via typefully.com.

U-Pool is a desktop provider and account switcher for Claude Code, Claude Desktop, Codex, Hermes, OpenCode, and Cursor. It keeps endpoint URLs, API keys, models, and provider accounts in per-tool lists, then makes a selected provider or Cursor account active without manual edits to settings.json, config.toml, other tool configuration files, or Windows user environment variables. Its reachability probe sends a plain GET request to a provider’s models endpoint to measure latency without running a completion or consuming a token. The app uses a Python backend, a statically exported Next.js interface, and a native OS webview rather than Electron or a Node runtime. Its source of truth is ~/.u-pool/config.json; on each switch, an adapter changes only the keys it owns in each target file and preserves other settings such as plugins, hooks, marketplaces, themes, MCP servers, and project trust. Writes use a temporary file followed by os.replace, keep an adjacent backup, and stop rather than overwrite a configuration file that cannot be parsed. Existing provider settings can be imported on first run. The app includes provider presets, an official-vendor mode that clears U-Pool-managed settings, per-provider permission controls, cached installed Claude Code and Codex versions, and controls that erase the transcript and prompt-history files for those CLIs. The Cursor pool stores multiple signed-in Cursor accounts, accepts cookies in common formats, displays account names, email addresses, plans, and usage, and switches accounts by changing the specified Cursor authentication database rows. Claude Desktop is currently preview-only: providers can be saved and tested but are not written to its configuration. Hermes and OpenCode have dedicated configuration writers, while Codex writes its provider, model, approval, sandbox, web-search, and authentication settings. The core configuration operations are cross-platform. Windows-only integrations write owned variables to HKCU\\Environment and broadcast changes, launch the app at sign-in, and provide an in-app updater; Hermes, OpenCode, and Cursor read credentials from their own files and do not use those environment-variable writes.

VidMap is an offline structure-from-motion system for video, developed by the Computer Vision and Geometry group (cvg). It combines temporal feature tracks, loop closures, metric depth, and global optimization to estimate camera poses, camera intrinsics, and a sparse 3D map from video or extracted frames. The Python package writes the reconstruction as a COLMAP model and supports optional browser-based HTML visualization and Rerun playback or flythrough recordings. The repository provides setup and execution workflows built around COLMAP, PyCOLMAP, PyTorch, xFormers, and GPU acceleration through CUDA and cuDSS.

WARP (Weight-Aware Runtime and Paging), formerly WASTE, is an embeddable inference engine written in dependency-free C by SQLiteAI. It runs large mixture-of-experts models beyond available RAM by keeping the shared model trunk in memory, streaming selected experts from NVMe or other fast storage, and using remaining RAM as a bounded expert cache. Its container layout gives each expert an aligned read; reads overlap computation, while a lookahead router prefetches experts for the next layer without changing the decisions made by the actual router. Experts use 3-bit residual vector quantization, while more sensitive shared weights use 4- or 8-bit storage. The project is designed for local inference of frontier-scale models such as Kimi K3 and GLM-5.3-Flash on consumer hardware.