2,421 tools and products — trending open source, and what gets used in AI and other work.
withoutBG is a Python SDK and command-line tool for removing image backgrounds locally or through a cloud API. Its local mode runs open-weight CPU ONNX models on the user's machine, downloading the model weights once; processing then works offline without an API key or GPU. The same API supports single-image and batch processing, progress callbacks, and returns transparent RGBA PIL Images. The package also provides a cloud mode through the withoutBG API and can be installed with uv or pip.
Agentic Inbox is a self-hosted email client from Cloudflare that runs on Cloudflare Workers through the user's Cloudflare account. It receives and sends mail through Cloudflare Email Routing, isolates each mailbox in a Durable Object with SQLite storage, and stores attachments in R2. The web interface supports threaded replies and forwards, rich-text composition, folders, search, and attachments. Its built-in AI agent, using the Cloudflare Agents SDK and Workers AI, can read inboxes, search conversations, draft replies, and send messages through email tools; automatic drafts for incoming mail require explicit confirmation before sending. Deployment provisions Workers, Durable Objects, R2, and Workers AI, and the setup requires Cloudflare Access, Email Routing, Email Service, and a mailbox on the configured domain.
Sockudo is a self-hosted, high-performance Rust real-time server for WebSocket and HTTP publishing. It targets strict Pusher Protocol V1 compatibility for drop-in use, while Protocol V2 adds Sockudo-native events, serials, message IDs, recovery, rewind, tags, deltas, annotations, and mutable message events. The platform supports durable history, hot replay buffers, two-tier recovery, presence history, push notifications, webhooks, rate limiting, TLS, Prometheus metrics, Docker Compose, and Helm. Horizontal fanout can use Redis, Redis Cluster, NATS, RabbitMQ, Google Pub/Sub, Kafka, Pulsar, or Apache Iggy. The monorepo includes the Rust server and workspace crates, dashboards, deployment assets, documentation, official real-time and server SDKs, and an optional AI Transport SDK.
OpenReply is an open-source, self-hosted Instagram comment-to-DM automation tool and ManyChat alternative. It watches Instagram post comments through Meta's official API, matches configured whole-word or partial keywords, and sends a private reply containing a link; it can also post a public reply under the comment. The same campaign keywords can trigger replies to inbound direct messages, including Story replies. It supports tracked redirect links with click and CTR statistics, up to two link buttons per DM, optional follow gating, username personalization, multiple professional Instagram accounts, workspaces and roles, and campaign templates. Requests are queued to remain within Meta's documented private-reply rate limit. A working installation requires the user's own deployment, domain, Meta app, API keys, and registered webhook; the project's public dashboard is a demo rather than a hosted service. OpenReply does not scrape Instagram, automate a browser, or request an Instagram password.
Mindwtr is a free, open-source, local-first task and idea management app built around the Getting Things Done (GTD) workflow. It captures tasks, ideas, and worries in an Inbox through desktop hotkeys, mobile widgets, share sheets, or voice input; guides users through sorting items into scheduled tasks, next actions, waiting-for items, projects, and Someday/Maybe; and provides Focus and guided weekly-review views. It supports projects with sections, areas, and manual task ordering, and is available for desktop, mobile, and web use. Data is stored on the device by default, with optional synchronization through iCloud, Dropbox, shared folders, a user-owned server, or WebDAV; the project states that it requires no account or subscription.
netmon is a self-hosted local-network monitor that periodically measures download and upload speed, ping latency, ISP, and test-server details, then scans the local subnet with nmap ARP scanning to count active devices. It stores the measurements and device counts in a local SQLite database and uses matplotlib to produce 24-hour trend graphs. The monitor sends status updates and periodic reports to Telegram or Discord. Reports can include sarcastic LLM-generated analysis through an OpenAI-compatible API, including a local LLM server. It runs on macOS or Linux, uses Python and uv, and requires system installations of nmap and speedtest-cli; device counting requires passwordless sudo access for nmap.
OpenObserve is an open-source, self-hosted observability platform for logs, metrics, traces, dashboards, frontend monitoring, pipelines, and incidents. It accepts OpenTelemetry data and provides SQL queries for logs and traces and SQL or PromQL queries for metrics. The platform is built in Rust and supports single-binary deployment, using Parquet columnar storage with an S3-native architecture; it also offers a unified web interface and cloud-hosted edition.
DVinyl is a self-hostable collection-management platform for cataloging and organizing physical media and collectibles. Its plugin-based design supports music formats such as vinyl, CDs, and cassettes; books; movies; video games; LEGO sets; and custom collection types. Built-in plugins retrieve metadata and cover art from Discogs, Hardcover, TMDB, IGDB, and Rebrickable, while items can be added by identifier or barcode, or imported in bulk from sources such as a Discogs library. It supports multiple collections, wishlists, music market-value estimates, configurable dashboards, themes, mobile-responsive views, multilingual use, user roles, OIDC single sign-on, and read-only public collection links or QR codes. Custom types can be created through a no-code plugin editor or added as code plugins through the API.
CrawlSEO is an open-source, self-hosted SEO monitoring dashboard for founders. It combines Google Search Console analytics for keywords, pages, clicks, impressions, click-through rate, and position changes with 28-day comparisons; a concurrent site crawler that handles up to 2,000 pages, assigns a health score, identifies 16 issue types, scores content, and provides per-page audit and remediation data; Core Web Vitals monitoring through PageSpeed Insights; historical rank tracking; SEO opportunity detection; alerts; and CSV export. Optional keyword research and backlink analysis use DataForSEO with a bring-your-own-key setup, while Google Autocomplete provides a free fallback for keyword suggestions. The included Model Context Protocol server exposes 10 tools for Claude Code, Claude Desktop, and Cursor. Connected AI agents can query SEO data, run crawls, inspect crawl issues and performance, and find opportunities. The application is built with Next.js, TypeScript, PostgreSQL, Prisma, and NextAuth.js, and is distributed under the MIT license for local or Docker Compose deployment with Google OAuth and a database. It is described as free and open source, with self-hosting intended to keep site data under the operator's control.
ShelfDroid is a third-party Android client for self-hosted Audiobookshelf servers. It lets users browse audiobook and podcast libraries, stream books and episodes with synchronized listening progress, and download them for offline playback. The app includes chapters, bookmarks, a sleep timer, playback-speed controls, and other player functions; accounts with permission can also manage backups, API keys, users, logs, server settings, sessions, notifications, authentication, metadata, and other server features. ShelfDroid is open source under the GNU Affero General Public License v3.0.
Parker is a self-hosted comic media server for organizing and reading CBZ and CBR files, using the filesystem as the source of truth. It parses metadata from ComicInfo.xml archives and organizes a library into series, volumes, and comics, with support for collections, story arcs, stacks, reading lists, and smart lists. Its reader supports manga right-to-left mode, double-page spreads, long view, bookmarks, navigation from series and lists, and per-comic reading settings. The server also provides multi-library access controls, metadata and storage reports, duplicate and corruption detection, background scanning, and optional parallel thumbnail generation.
Hopp is an open-source remote pair-programming and screen-sharing application for developers. Its native desktop client uses Tauri with Rust and React/TypeScript, while a Go backend uses PostgreSQL and Redis; screen sharing runs over WebRTC infrastructure powered by LiveKit. The application supports one-click pairing, rooms for up to 10 participants, and the project claims sub-100ms latency. It can be used through Hopp's managed cloud or self-hosted with Docker Compose, which provisions PostgreSQL, Redis, LiveKit, and Caddy. macOS is supported with a stable build, Windows with an alpha build, and Linux is planned.
compose-lint is a security-focused static-analysis linter for Docker Compose files such as docker-compose.yml and compose.yaml. It checks for privileged containers, missing capability restrictions, host namespace or network sharing, wildcard port bindings, unpinned images, Docker socket and sensitive host-path mounts, plaintext credentials, and related misconfigurations, with rules grounded in the OWASP Docker Security Cheat Sheet and CIS Docker Benchmark. The Python tool runs locally or in CI, can export SARIF results, and provides mechanical fixes for unambiguous findings with a dry-run preview. It is distributed through PyPI, can also be run ad hoc with uvx or pipx, and has a multi-architecture Docker image that runs as a non-root user.
PeaNUT is a web dashboard for Network UPS Tools (NUT) that monitors network-connected UPS devices. It displays real-time battery, runtime, load, and status information, supports multiple devices and UPS commands, and provides configuration through a web UI or YAML file. It offers an API, Prometheus support, InfluxDB v2 integration for Grafana monitoring, and customizable widgets for Homepage integration. The project is distributed as a Docker image and supports Linux on amd64 and arm64 platforms, including Raspberry Pi 4 and newer.
Dockge is a self-hosted web manager for Docker Compose stacks. It manages compose.yaml files without taking ownership of them: users can create, edit, start, stop, restart, delete, and update stacks, while continuing to use the files with normal Docker Compose commands. Its interface includes an interactive Compose editor, a web terminal, and real-time pull, startup, shutdown, and terminal output. Multiple-agent support allows stacks on different Docker hosts to be managed from one interface, and a converter turns docker run commands into compose.yaml files. Dockge runs with Docker 20 or newer or Podman on supported Linux distributions, and the repository documents Docker-based installation with a default port of 5001.
CheapSecurity is a lightweight, self-hosted CCTV application for Linux-based single-board computers such as Raspberry Pi and standard USB webcams. It provides a Flask web dashboard with live MJPEG streaming, frame-difference motion detection, pre-motion buffering, automatic recording, a recordings calendar, and storage cleanup based on age, total size, or low disk space. The application can send motion snapshots by email and recordings or snapshots through Telegram, Google Drive, or OneDrive. It supports optional AES-256-encrypted ZIP uploads, software night-mode enhancement using CLAHE and brightness/contrast adjustment, an interactive REST API with Swagger UI, optional RTSP output through MediaMTX and FFmpeg, and systemd autostart. It requires Python 3.10 or newer and OpenCV with V4L2 support, and is licensed under GNU AGPLv3.
Langfuse is an open-source LLM engineering platform for developing, monitoring, evaluating, and debugging AI applications. It ingests traces for LLM calls and related operations such as retrieval, embeddings, and agent actions, with integrations including OpenTelemetry, LangChain, the OpenAI SDK, LiteLLM, and other LLM frameworks. Its features include prompt management with version control and collaborative iteration, LLM-as-a-judge and code-based evaluations, user feedback and manual labeling, datasets for test sets and benchmarks, trace replay and debugging, metrics, and a playground for testing prompts and model configurations. Langfuse exposes APIs and typed Python and JavaScript/TypeScript SDKs for custom LLMOps workflows. It is available as a managed cloud service or for self-hosting with Docker Compose, virtual machines, Kubernetes via Helm, or Terraform templates.
cairn is a self-hosted directory page for people who use services hosted by someone else, rather than an administration dashboard. It presents tools grouped by purpose with welcome notes, one-sentence descriptions, optional detail pages, service-status indicators linked to status pages, lifecycle labels, browser-language selection with a manual switcher, and search across names, descriptions, and tags. Its server performs status polling so visitors' browsers do not, and features degrade without JavaScript. Operators configure it with a read-only YAML file that applies changes live and reports configuration errors with the file, line, and expected shape. It is distributed as a static binary with no runtime dependencies or database, can serve legal pages written in Markdown, and supports deployment on a domain, subdomain, or sub-path. The interface includes responsive, keyboard-friendly, multilingual, right-to-left, and stated WCAG AA-oriented accessibility features.
Expense Budget Tracker is a self-hosted, open-source personal-finance application for budgeting, tracking transactions and balances, transferring funds between accounts, and reporting across currencies. It runs on PostgreSQL and provides a web interface for managing budgets, transactions, accounts, and workspaces. The project includes a hosted MCP server and an Agent API. Compatible MCP clients can inspect workspace schemas, run restricted read queries, and, with the required write scope, execute approved data changes. The HTTP SQL API uses API keys and a flat table structure so AI agents can query and analyze financial data or import transactions from screenshots, CSV files, and PDF bank statements. The repository provides a Docker-based quick start that runs PostgreSQL, migrations, the web application, and a worker. The project is fully open source and designed for deployment on the user's own servers.
fak is a self-hosted agent kernel distributed as a single Go binary. It wraps existing coding-agent sessions without replacing their interface or model, managing shared context, provider cache continuity, session recovery, model routing, tools, policies, and execution receipts. Tool calls pass through explicit policies that can allow, deny, transform, or witness them. Its default-deny capability floor blocks tools outside the configured allow-list, while the system can forward an existing provider subscription credential, route tasks to a configured provider or its native inference engine, and avoid silent fallbacks. The repository also provides an offline proof mode and installation through a shell script or Go.
getbased is an open-source, local-first personal health dashboard and self-hostable web application, with a hosted version at getbased.health. It combines laboratory results, DNA data, wearable and body metrics, light exposure, lifestyle context, therapy histories, and personal notes in a browser-based workspace. Lab reports can be imported from PDFs, images, and spreadsheets or entered manually; known markers are mapped for review, then tracked with charts and context-aware ranges. The application calculates supported ratios and indices through deterministic fallbacks, including biological age when the required inputs are available, and produces Biology Scores that summarize marker patterns rather than diagnose conditions. DNA support includes consumer and clinical raw-DNA formats, curated SNP interpretation, APOE haplotypes, and mitochondrial DNA haplogroups. The dashboard also supports selected wearable connections, local Apple Health file imports, manual measurements, light and environmental tracking, and structured therapy histories. It can run without an account and stores most data in the browser by default; AI analysis, encrypted synchronization, profile sharing, external Knowledge Bases, and Agent Access are opt-in. The project can be self-hosted or used through its hosted application.
backup-maker is a self-hosted, local-network backup application that mirrors selected folders in real time to multiple destinations, including internal or external drives, SD cards, router-attached USB storage, NAS devices, and shared folders on other computers. It installs only on the computer being backed up; destination systems require no software or accounts. Backups are incremental and versioned, with file history and scheduled AES-256-encrypted snapshots. An optional second backup-maker installation enables block-level, verified delta synchronization between computers running different operating systems. The project is distributed as a self-contained binary for Linux, Windows, and macOS, with no cloud service, off-site mode, telemetry, or required internet connection. Its dashboard listens on localhost, while update checks and webhook or ntfy alerts are optional.
Superlog is an open-core, self-hosted observability workspace for OpenTelemetry data, developed by Superlog Labs. It ingests traces, logs, and metrics, groups noisy signals into incidents, and provides a local-first interface for investigating production systems. The open-source community edition includes a web app and API, an OTLP intake proxy, worker processes for incident grouping and background jobs, a Postgres schema, ClickHouse-backed telemetry queries, and pluggable agent-runner interfaces with a default runner that records a local incident summary. The repository is licensed under Apache License 2.0; a hosted Superlog Cloud edition is also available with a free tier, pay-as-you-go plan, and monthly credit packs.
Renfield is a self-hosted, voice-first AI household assistant built with FastAPI, React, and Ollama. It uses Raspberry Pi Zero 2 W voice satellites with ReSpeaker HATs for local wake-word detection, Whisper speech recognition, Piper text-to-speech, and SpeechBrain speaker recognition. The system can track room presence through BLE scanning, voice recognition, and web authentication, and integrates with Home Assistant and DLNA for multi-room control. Its ReAct agent chains tools through 10 MCP servers covering weather, web search, news, calendars, Jellyfin, DLNA, n8n workflows, Home Assistant, Paperless, email, and related services. It also provides conversational long-term memory with contradiction detection, a knowledge graph with validated entity-relation triples, and a RAG knowledge base that combines dense pgvector embeddings with BM25 full-text search using reciprocal-rank fusion. Documents can be uploaded from formats including PDF, DOCX, PPTX, XLSX, HTML, Markdown, and TXT, with Tesseract or EasyOCR available for scanned or poor-quality PDFs. The project is designed to run entirely on local hardware without cloud dependencies and includes a web interface, satellite monitoring, proactive notifications, plugin hooks, and Paperless document-audit workflows. The repository describes more than 100 available tools and supports Docker-based self-hosting.
News Dashboard is a self-hosted technical-news inbox and AI research desk that aggregates curated RSS and Atom feeds, GitHub release feeds, Hacker News and GitHub trending feeds, and custom scraped sources into curated feeds and a ranked Today Feed. Articles can be marked new, read, saved, skipped, archived, starred, or snoozed; the application also provides source-health monitoring, ingestion history, dashboard statistics, search, reading history, a knowledge graph, and saved learning artifacts. When an AI provider is configured, it generates personalized briefings, article takeaways, context, and perspectives, and answers cited follow-up questions grounded in the collected corpus. It supports embeddings, configurable OpenAI-compatible providers, and a Google Reader-compatible synchronization API for third-party RSS clients. The application uses a Python and FastAPI backend with APScheduler, LangChain, and LangGraph; a React, TypeScript, Vite, and TanStack Query frontend; and PostgreSQL with the pgvector extension. It supports local password authentication, optional Keycloak login, and deployment through Docker, Helm, and GitHub Actions on the operator’s own infrastructure, with configuration for the database, authentication, and AI provider.
GodotHub is a Godot Engine project manager that combines project organization, Godot version management, reusable templates, assets, and Git operations in one application. It can create or import projects, clone them from Git repositories, organize them with categories, tags, pins, search, filtering, and drag-and-drop ordering, and show project properties such as file-type breakdowns. Its Git features include repository management and diffs, while its version manager fetches official Godot builds, supports importing existing installations, and provides resumable downloads with configurable concurrent transfers. The README identifies project and version management, templates, an asset store, Git integration, customizable appearance, and a Godot community news feed as core features. The application has been tested on Windows, Arch Linux with Hyprland, Fedora 43 with GNOME, and a recent macOS version; the README warns that behavior on other Linux distributions is not guaranteed and that AppImage installation is currently affected by an upstream linuxdeploy issue.
PGSimCity is an independent, non-commercial educational visualization of PostgreSQL internals: an explorable 3D city that represents a PostgreSQL cluster and lets users inspect, walk through, and disrupt simulated database scenarios. It illustrates effects such as checkpoint-related latency spikes, long-running transactions that keep tables bloated, and the cost of synchronous_commit. The city is a scaled model rather than a PostgreSQL emulator; PostgreSQL source code does not run in the 3D scene, and its quantities are adjusted for visualization. Its optional Query flow and Machine can run PGlite, an in-memory PostgreSQL compiled to WebAssembly. The project targets PostgreSQL 18 and provides keyboard- and text-first routes for its lessons, while noting that the first-person 3D scene has no complete nonvisual equivalent.
ReScript is an open-source, transcript-based video and audio editor that runs in the browser or as a desktop application. It processes media locally on the device, transcribes audio with Whisper using word-level timestamps, and assigns speaker labels with pyannote; users edit the media by deleting words from the transcript, which creates cut ranges that the preview and timeline apply to the original recording. The editor supports imported SRT, VTT, and JSON transcripts, filler-word and silence removal, speaker-grouped transcripts, waveform and word-level timeline editing, split and trim operations, and adjustable word timings. It exports video, audio, transcripts, subtitles, and NLE timelines for DaVinci Resolve, Premiere, Final Cut Pro, Pro Tools, and Logic. The web and desktop versions use ffmpeg.wasm for on-device media extraction and export. Desktop builds are distributed for macOS, Windows, and Linux through Electron, while the browser app is available at app.getrescript.com.
best-claude-hud is a Rust-based statusline HUD for Claude Code terminals. It displays the active Claude model and live reasoning effort, Claude Code's launch directory, workspace, Git branch and status including ahead/behind counts, and context-window usage from Claude Code's statusLine data with an active-transcript fallback. Optional segments provide usage or rate-limit metadata, cost, session, and output-style information. It is distributed through npm with prebuilt native binaries, so Rust is not required for installation. The repository also provides a Nix flake for running or installing the tool in declarative environments. Its setup command writes the required statusLine configuration to Claude Code's settings file while preserving existing settings.
agentacct is a local-first audit and observability tool for coding-agent work. It reads session logs from Claude Code, Codex, OpenCode, and Hermes, combines them with recorded tasks and test evidence, and produces a Work Receipt for each task covering commands run, files touched, tools used, time and token usage, cost, outcome, and evidence strength. Agent claims and machine verification are kept as separate axes, with verified status reserved for tasks whose live checks pass after the latest recorded work. The data is available through a macOS app, the live terminal dashboard launched with `agentacct tui`, and a local JSON API. State is stored in local files; the API listens only on loopback, and the project states that it has no login, cloud sync, or telemetry and does not request provider API keys.
AI Copywriter is a portable, plain-Markdown agent skill for writing marketing copy and removing common patterns associated with AI-generated prose. It supports headlines, short descriptions, microcopy, subject lines, button labels, and LinkedIn posts, and runs in any harness that supports skill-style instructions. Before drafting, it interviews the user about the intended customer, product category, and underlying story, then pressure-tests the story for specific and interesting details. Its copywriting method focuses on the reader's immediate situation and explains the product in simple language. It incorporates the 33 detectable and fixable writing patterns from blader's Humanizer while adding the reverse workflow of generating marketing copy designed around those constraints.
MoonEP is an expert-parallelism communication library from Moonshot AI for mixture-of-experts training and inference. It plans a small number of dynamic redundant experts from the current router outputs, prefetches their weights before expert computation, and reduces the resulting gradients back to the experts' home ranks during the backward pass. Its near-optimal GPU planning kernel, fused permute/unpermute operations, and zero-copy dispatch send tokens directly to expert-grouped positions on remote ranks and return buffer views to computation. The library uses a fixed S × K token buffer per rank, where S is the input-token count and K is the routed top-k, keeping communication and computation shapes static across layers. The repository documents support for NVIDIA GPUs and lists Zhenwu PPU support as under review.
Crisp is a free, open-source native menu bar app for controlling external monitors on macOS. It adds sharp HiDPI scaled resolutions, fine-grained display scaling, hardware brightness and volume control over DDC, software dimming for unsupported monitors, display presets, display arrangement, color and system toggles, and virtual displays. Presets can store named resolution, brightness, and arrangement configurations, while the app can disconnect and reconnect physical displays. Crisp is distributed for free without a Pro tier or license key. It can be installed through Homebrew or a downloadable DMG; the repository says releases are signed and notarized by Apple. The app is localized in English and Simplified Chinese.
Better Harness is an open-source Harness Engineering platform for coding agents. It analyzes project and, where supported, session evidence around an agent's work rather than only its final code diff, evaluating the Agent Work Loop for issues such as task understanding, validation, delivery control, and retained lessons. It turns supported gaps into prioritized findings linked to their evidence, expected outcomes, repair boundaries, and acceptance checks, and produces host-specific reports in formats including HTML, paired Markdown, or native Canvas reports. The platform also supports defining harnesses as code, running controlled experiments, inspecting evidence, and comparing outcomes across coding-agent hosts such as Claude Code, Codex, Qoder, Cursor, and GitHub Copilot CLI.
humanizer-stack is an open-source writing-humanization toolkit packaged as Claude Code Skills. Its two-pass pipeline first scans and revises surface-level tells such as hype vocabulary, em-dash overuse, formulaic phrasing, antithesis, vague attribution, inflated symbolism, and redundant exposition; a second structural pass audits discourse-level patterns grounded in the StoryScope study. The structural audits cover theme explicitness, structural tidiness, emotional framing, reference specificity, reader engagement, and shape convergence, with deterministic Python scanners for mechanical copy and structural tells. The repository includes the skills, reference materials, genre-specific guidance, and documentation describing how the passes are chained.
Lifeline is a timeline component for presenting stories that unfold over time, such as a career, company, or journey. It lays milestones on a single rail that is horizontally scrubbed by scrolling on desktop and becomes vertical on mobile; milestones can contain years, events, links, people, images, and video, with media shown as hover reveals or floating cards that expand into a lightbox. An introductory animation draws the rail across the years. The project ships as a shadcn registry: installation copies the components and supporting data helpers into an application's codebase, where their styling, breakpoints, and animation behavior can be modified. It provides page, shell, starter-data, component-system, and theme-switcher registry items for Next.js projects.
Moumen Lab is a React component registry for interface components such as menus, OTP inputs, upload staging, command palettes, and draggable reorderable lists. Components are distributed in the shadcn registry format and installed as source code with the `moumenlab` CLI or shadcn CLI, so no runtime package is required. The components use Tailwind CSS with shared `lab-theme` tokens and motion-based animations that are interruptible, layout-aware, and reduced-motion aware. The repository includes live showcases, component blueprints, source files, registry metadata, and a monorepo-based CLI; it targets React 19, Tailwind CSS v4, Node.js 18+ for installation, and is licensed under MIT.
Lexicon is a free, open-source, local-first writing assistant and distraction-free rich-text editor for Windows, macOS, and Linux. It provides inline proofreading through the separate LanguageTool engine, while local AI tools handle rewriting, tone adjustment, summarization, and other writing tasks without requiring an account or cloud writing service. Users can review suggestions in a dedicated panel, format drafts with rich text and LaTeX, autosave work, and export it to Markdown, HTML, text, or PDF. The app can download and run a local AI model or use an existing local Ollama server; the editor and proofreading tools work without an AI-model download. It is MIT-licensed, and the repository notes that its Windows builds are unsigned and its macOS build is pre-release and unsigned.
undo is a Linux shell tool that reverts the filesystem changes made by the last shell command without rolling back unrelated work. It uses shell hooks for Bash, Zsh, and Fish to arm an LD_PRELOAD library around each command; the library intercepts destructive filesystem operations, including deletion, renaming, writes, directory creation, and permission changes, saves affected files in a per-command session, and records the changes in a journal. The tool replays the journal in reverse to restore deleted or overwritten files, reverse renames, remove accidental creations, and restore directory modes. Deleted files are preserved with hard links, and the session store is made of inspectable plain files. It provides undo and redo operations, keeps a configurable set of sessions within a configurable storage budget, requires no daemon or root privileges, and supports a purge operation for clearing the store.
Shell Gym is an interactive Linux command-line trainer implemented as a background daemon with a built-in web UI. It presents short assignments, or “reps,” such as entering a directory, creating a file, killing a process, or freeing a port; the assignment completes automatically when the corresponding system state changes. Users type real commands in an ordinary terminal while the UI observes the host and provides guidance, without modifying the shell or using prompt hooks, wrappers, special functions, a check button, or copy-and-paste. Its outside-the-shell observation uses procfs to discover interactive shells and inspect working directories, processes, files, and ports; the Linux kernel proc connector to observe exec() events; and direct state checks for files, processes, and listening ports. It runs on a Linux host as root, with an online playground also available, and the project advises using a disposable VM or other disposable Linux system because exercises perform real actions on the live host.
opentax-engine is an open-source deterministic US tax calculator designed for AI agents and other applications. It accepts tax facts such as wages, filing status, and dependents, applies an encoded rule corpus, and returns cent-exact results with assumptions, legal citations, and a proof tree showing how each value was derived. If its rules cannot derive an answer, the project is designed to refuse rather than guess. The calculation can be selected for a date with an --as-of option, and the CLI accepts flags or JSON fact files; the videos also identify CLI, browser, and MCP integrations. The engine is implemented in TypeScript with no platform dependencies and can run in a self-contained browser HTML file containing the engine, rule corpus, and verifier. Proofs use canonical JSON, hashing, and a Merkle construction; the verifier independently re-derives the calculation and reports altered proof data, rule-corpus differences, or steps that do not match. The repository describes the project as AGPL-3.0 licensed, with commercial licenses available.
Audio-Cassette-Simulation is a set of local Bash scripts that uses FFmpeg to simulate cassette-tape audio profiles. It applies tape noise, wow and flutter pitch modulation, bandwidth limits, and equalizer adjustments through presets for BASF LH Extra C90, Maxell UD C90, Sony CHF60 and CHF90, TDK D90, and Soviet MK-60 tape. The scripts convert local MP3, OGG, M4A, and AAC files, or record a live web stream into segmented files, saving processed output in an `out` directory. It requires FFmpeg compiled with libmp3lame and is licensed under the MIT License.
Yap is a free, open-source voice-dictation app for macOS developed by Frigade. It runs in the menu bar, uses macOS 26's SpeechAnalyzer and SpeechTranscriber APIs for on-device streaming speech-to-text, and ships no transcription model of its own. A keyboard shortcut starts and stops recording; the resulting text is pasted into the active text field, while transcripts are saved locally for later copying. It requires an Apple Silicon Mac running macOS 26, uses no account, API key, or network connection, and is distributed as a signed and notarized app, a disk image, or a Homebrew cask.
Amber Console is a monochrome amber-terminal CSS framework for web interfaces styled like late-1980s industrial control panels. It uses brightness, inverse video, and blinking rather than hue for hierarchy; 4px half-cell spacing based on a character grid; and stroked borders instead of elevation. The framework includes framed controls, instrument-style readouts, multiple palettes, and optional CRT-like effects such as ghosting, scroll smear, framebuffer decay, residual patches, and phosphor afterimages. It is distributed as a stylesheet with no dependencies, build step, or JavaScript required for component appearance; optional JavaScript modules add behavior and persistence effects. The project provides CSS and JavaScript distribution builds and can be installed with a link tag or imported as a package.
The Long Silence is a browser-based procedural space-exploration game. It generates stars, planets, ring systems, nebulae, and derelicts from a seed rather than using downloadable art assets; the player pilots the Pale Seeker, charts systems, scans discoveries, lands on worlds, and attunes to Resonators across the game’s seven-system objective. It uses WebGL 2, handwritten GLSL, a floating origin, and a logarithmic depth buffer to represent systems spanning millions of world units while keeping the ship near the origin. Planets are baked into cubemaps containing albedo and terrain-height data, with runtime shaders sampling those textures for normals and lighting. Atmospheres use single-scattering raymarches, while GPU auto-exposure and a hand-built post-processing chain provide bloom, anamorphic streaks, god rays, lens ghosts, tone mapping, grain, dithering, and FXAA. The project is developed as a JavaScript package with npm scripts for local development and production builds; the latter produces a static bundle in the dist/ directory. Its renderer adjusts resolution dynamically between 0.62× and 2× to target frame rate rather than removing visual features.
τ₀-VLA is an open-source hierarchical robot foundation model and reference implementation for long-horizon manipulation. A memory-augmented high-level policy generates the next subtask and uses world-model-guided test-time computation to search over alternatives when additional reasoning is needed; a generalist low-level policy then executes the selected subtask across robot embodiments. The low-level policy combines a Qwen3.5 vision-language backbone with a Mixture-of-Transformers action expert trained through conditional flow matching, using a unified 40-dimensional state/action space and multimodal co-training on tens of thousands of hours of heterogeneous real-world robot data. The repository provides LeRobot-format data loading, prompting, masking and normalization, embodiment-specific adapters, training and post-training recipes, deployment contracts, a policy server, and open-loop evaluation tools. It includes an AgiBot World example subset and templates for other datasets and robots. Public v1 serving supports joint-control checkpoints; native end-effector data can be used for training but end-effector serving is not supported in that release. The reference environment uses Python 3.11, CUDA 12.8, and PyTorch 2.7.1. Code and model weights are released under the Apache License 2.0.
OptMem is a permanent-memory tool for AI agents. It installs as a dependency-free Python 3 script and integrates through a prompt block added to an agent's AGENTS.md or CLAUDE.md file. The agent uses `memo wake` at session startup, `memo note` to append one-line memories, `memo recall` for exact regular-expression searches, and `memo zoom` to navigate summaries. Raw memories are stored in an append-only `LOG.txt`; a binary tree of one-line summaries provides a compact reading view, with summaries rebuilt when needed. `memo nap` processes due merges, while `memo forget` discards a summary so the next nap can rebuild it. Memory can be kept in a configurable directory such as a synced folder or Git repository, and the system is designed to persist across sessions, context compaction, models, and vendors.
Herdr Browser is a deprecated browser-automation experiment that renders a real Chromium view inside a Herdr terminal pane. It captures Chromium frames through the Chrome DevTools Protocol, encodes them as PNG images, and sends them through Herdr's pane graphics stream for display in the terminal; automation clients can drive the browser over CDP while users view or interact with the persistent session. The project is no longer maintained and directs users to Terminal Browser, which uses Electron offscreen rendering and file-backed raw RGBA frames instead of the PNG encode/decode path.
Shiki (私記) is a Rust terminal user-interface note-taking application. It organizes Markdown notes into notebooks, with each notebook stored as an independent Git repository, and provides a responsive three-pane, Yazi-style navigation layout with nested folders, optional frontmatter, inline or external editing, themes, and global or notebook-scoped fuzzy search. Its per-note history is derived from Git commits: users can browse commits that changed a specific note and revert from within the TUI. The application also supports manual or automatic notebook synchronization and push/pull operations, configurable per-notebook paths and policies, remappable keybindings, mouse selection in rendered previews with OSC 52 clipboard copying, and live theme previews. The repository documents 37 built-in themes and support for using existing note directories, including an Obsidian vault, without migration or duplicated files.
Deltafin is a native binary for running the full, unpruned Kimi K3 mixture-of-experts model on consumer hardware, along with an OpenAI-compatible API server for local chat and coding agents. Its demonstration approach keeps the model's attention spine on local disk and fetches routed experts from Hugging Face's CDN as needed, allowing the full model to run on a 64-gigabyte MacBook rather than loading the entire expert bank into memory. The project emphasizes preserving Kimi K3's original expert weights and having K3 verify every token; it is presented as an experiment in pushing large-model inference on local hardware rather than as a practical chat setup.
Druk is a terminal-based code editor distributed as a self-contained executable, with no Node or Bun runtime required. It provides a file tree, tabs, project-wide fuzzy file search, syntax highlighting for more than 30 languages using tree-sitter, language-server support, Git marks, extensions, PDF viewing, and keyboard and mouse controls. Binaries are provided for macOS, Linux, and Windows, with installation options including a shell script, Homebrew, npm, Bun, Debian, and RPM packages.
OpenWorker is an open-source desktop AI coworker that produces finished deliverables from everyday tasks, including code-security reviews with proposed fixes, cloud-configuration audits, incident reports, documents, spreadsheets, drafted messages, and scheduled briefs. It runs on the user's machine through a native desktop app and a local Python agent server built on aisuite, working across local files, the terminal, connected applications, and more than 25 connectors. Specialist coworkers cover security review, cloud posture, incident triage, everyday work, and recurring automations; security review combines deterministic scanners such as Semgrep with model reasoning, then re-scans and diff-reviews proposed fixes before approval. The agent breaks requested outcomes into steps and uses models from providers such as OpenAI, Anthropic, and Google, open-weight providers, or a local Ollama deployment, with the user supplying model access. Consequential actions—including sending messages, changing calendars, and running commands—are approval-gated. Its governance design includes human-only floors for dangerous or irreversible operations, explicitly granted and revocable autonomy rules, circuit-breaker escalation for uncertain or repeatedly denied actions, and an audit trail recording tool calls, approval provenance, and reviewer reasoning; unattended runs cannot self-approve. The project is in open beta and provides downloads for macOS on Apple Silicon and Windows on x64.
scroll-world is an agent skill for Claude Code, Codex, and other SKILL.md-compatible agents that builds scroll-scrubbed 3D world landing pages for brands and industries. It generates cohesive isometric diorama scenes and image-to-video camera flights, linking scenes through first/last-frame conditioning so the camera appears to move continuously as the visitor scrolls. The skill provides prompt templates, an AI image and video rendering pipeline using Higgsfield, Monid, Seedance, Kling, or Codex image generation, and a framework-agnostic vanilla JavaScript scrubbing engine that can be used with plain HTML, Next.js, Vue, or a Python-served page. It can also render a separate portrait chain for mobile devices and requires external rendering services or CLIs, ffmpeg/ffprobe, and Python with Pillow.
An open, machine-readable taxonomy of learning across the primary and elementary years, produced by Marble. It represents fine-grained micro-topics as nodes in a prerequisite graph, with each topic containing a plain-language description, subject and domain, conceptual or procedural type, approximate age range, mastery evidence, assessment prompt, and links to source curriculum standards such as NGSS, Common Core, and the UK National Curriculum. Dependency edges identify hard or soft prerequisites and include reasons; the dataset also provides parent-friendly domain summaries. The JSON files include topics, dependencies, curriculum standards, domain clusters, and a manifest with counts and checksums. The repository describes it as pure data with no runtime or dependencies, and provides an interactive curriculum explorer at withmarble.com/curriculum.
No AI Slop is an AI-writing editing skill that removes more than 20 canned machine-writing patterns while preserving the author's vocabulary, cadence, humor, and imperfections. Its rules target binary contrasts, throat-clearing openers, faux-insight setups, colon reveals, dramatic fragments, superficial analysis, importance puffery, weasel attribution, synonym cycling, and fake-profound endings; they also cover active voice, leading with the point, untangling difficult sentences, and preferring concrete details over abstractions. The skill supports editing text, detecting and quoting suspected patterns without judging whether AI produced the writing, and generating deliberately exaggerated AI writing for satire. It can be invoked as `/no-ai-slop` in ChatGPT, Claude Code, Codex, or another coding agent, and is distributed through the Skills package manager or an npx command. The repository contains the editing rules in SKILL.md, evaluation checks in eval.md, ChatGPT and Codex plugin metadata, and a plugin build script; it is licensed under MIT.
OpenScience is an open-source AI workbench for scientific research developed by Synthetic Sciences. It runs as a browser-based workspace where a single research agent takes a goal through literature review, hypothesis formation, code writing and execution, experiments, analysis, and a written report. The agent can load domain skills, delegate bounded exploratory or execution work, query scientific databases, and preserve sessions as observable research traces. The local server hosts the workspace UI, agent runtime, skill library, and tool layer. The agent plans with a research harness and calls shell, editor, LSP, MCP, scientific database, and skill tools; sessions, skills, artifacts, and provenance are stored on disk. The workbench includes a file tree, code editor, terminal, session history, and inline rendering for molecules, structures, genomes, and plots. Its bundled skills cover training, evaluation, datasets, molecular and clinical biology, cheminformatics, papers and LaTeX, figures, and scientific runtimes, while database connectors provide direct access to services including UniProt, PDB, Ensembl, ChEMBL, PubChem, arXiv, OpenAlex, and Semantic Scholar. It supports frontier and open-weight models through user-provided keys, eligible ChatGPT or Codex access, local models, and optional Ace-managed models. Models are routed per request, allowing providers to be switched without changing the workspace. Extensibility includes LSP integration, MCP servers, plugins, custom agents and commands, experimental Python environments and BioNeMo adapters, and a TypeScript SDK. It is installed with the @synsci/openscience npm package or launched through npx; platform binaries and desktop installers are distributed through GitHub Releases. A free Synthetic Sciences account links an installation and can provide credential synchronization, private research graphs, enhanced search, and optional credit-backed models, while BYOK and local-model usage remain separate from Ace.
An open-source project that runs a 28.9-million-parameter language model locally on an ESP32-S3 microcontroller, without network connectivity. It uses a tiered memory layout based on Per-Layer Embeddings: frequently accessed activations and normalization weights stay in SRAM, the model core and output head reside in PSRAM, and a roughly 25-million-parameter embedding table is stored in flash; only the few rows needed for each token are read into faster memory. The included TinyStories model generates short stories, while a Barista model provides espresso question answering. The repository notes that the TinyStories model is not designed for general question answering, instruction following, code generation, or factual knowledge, and provides separate scripts for downloading and verifying model assets and deploying a selected model to the board.
Harness Engineering is Ryan Lopopolo’s anthology, field guide, and agent-context repository for improving coding-agent output without changing the underlying model or coding agent. It shapes the surrounding environment through context and tools so an agent can recover intent, operate the real system, respect authority, prove its outcome, and incorporate lessons from accepted work, corrections, failures, and user responses. The repository uses AGENTS.md to route agents to relevant arguments, cases, and proof, and provides a thesis index, playbooks, source material, and executable constraints for encoding organizational requirements and decisions. Repository-authored material is licensed under CC BY 4.0.
The Fable Method is an agent-workflow repository that distills the Fable Workflow into skills that models can run. Its process classifies the request, defines completion with named verification, gathers evidence from primary sources in parallel, commits to one recommendation, changes the smallest correct thing, verifies the result by observation, and reports the outcome with caveats. The repository provides four related skills—fable-method for thinking, fable-loop for acting, fable-judge for evaluating results, and fable-domain for generating domain adapters—along with evaluation cases, transcripts, judge outputs, and logs. The project reports testing the method across fifteen evaluation rounds and more than 260 agent runs, with judges checking diffs and execution rather than relying on agent reports.
riddle is a Rust application that turns a reMarkable Paper Pro into a diary-style AI interface. It reads pen strokes from the tablet, waits for an idle pause, commits the page as a PNG to a resident oracle LLM process, and streams the response back as animated handwriting. The handwriting pipeline rasterizes the response, applies Zhang-Suen thinning, traces the result into single-pixel pen paths, and replays those paths stroke by stroke. The app supports a qtfb display backend for running inside xochitl and a quill-based takeover backend that drives the vendor e-ink engine directly; the prebuilt bundle uses takeover mode. It can be configured with an OpenAI-compatible API key or used with pi, according to the repository instructions. Installation requires a reMarkable Paper Pro in developer mode with xovi and AppLoad, or the remagic installer. Takeover mode stops the normal reMarkable interface, runs as root, and takes control of the display; the repository says it has been tested on the Paper Pro ferrari model with OS 3.26–3.27 and warns that installation modifies the device and may not work on other models or versions.