High-resolution weather forecasts became a free commodity because of Open-Meteo - and this deployment puts the whole open-source engine on your own infrastructure. The public open-meteo.com service aggregates national weather models (NOAA GFS, DWD ICON, ECMWF, Meteo-France, and others) into one consistent JSON interface; self-hosting gives you that same API without rate limits, third-party dependency, or usage metering. The architecture is two cooperating services: the API server exposes forecast endpoints fully compatible with Open-Meteo query parameters - latitude, longitude, hourly and daily variables like temperature, precipitation, wind, and radiation - while a background sync worker downloads fresh weather model data on a configurable interval into a shared persistent volume at /app/data, so forecasts stay current and survive restarts without re-downloading. You control which weather models to mirror, which variables to store, how much historical depth to keep, and how often to refresh - meaning a lean deployment can sync only the model and region you actually query. Responses are plain HTTP/JSON, so integration with dashboards, Home Assistant-style automations, agricultural monitoring, IoT fleets, or any application takes minutes. For anyone making thousands of forecast calls a day, replacing a metered weather API with your own instance turns a recurring bill into a flat infrastructure cost.

Security-conscious IT departments pick Passbolt for its cryptography: every user holds an OpenPGP key pair, and shared credentials are encrypted individually to each recipient's public key - real end-to-end encryption, not a vault password handed around. All crypto runs client-side in the mandatory browser extension (distributed and signed through the Chrome and Firefox stores, deliberately separating the crypto code from the server that stores ciphertext); private keys and passphrases never touch your instance, and the server admin cannot read a single secret. Authentication uses the challenge-based GpgAuth protocol, secrets are digitally signed to verify sender integrity, and metadata encryption extends protection to resource names and URLs. Day to day it behaves like a polished commercial manager: auto-fill and auto-save in forms, strong password generation, anti-phishing protection, TOTP storage, folder hierarchies shared per-user or per-group with fine-grained permissions and instant cryptographic revocation. Native iOS, Android, and desktop apps ship alongside a JSON API, CLI, and SDKs for CI/CD secret retrieval and rotation. The PHP server runs on MariaDB and is AGPL-licensed open source - including the paid tiers' codebase - with published security audits.

What Jellyfin is for video, Komga is for comics, manga, BDs, magazines, and eBooks: point it at folders of CBZ, CBR, CBT, PDF, and EPUB files and it becomes a proper library with cover art, embedded metadata import, and per-user reading progress. The responsive web reader handles multiple reading modes - left-to-right for western comics, right-to-left for manga, webtoon scrolling - while collections and read lists organize crossovers and story arcs, with ComicRack .cbl read list import for existing curation. Its openness is the real differentiator: OPDS v1.2 and v2.0 feeds serve virtually every reader app (Panels, Moon+ Reader, Mihon/Tachiyomi, KyBook), with OpenSearch and page-streaming extensions so mobile apps fetch pages on demand instead of whole archives. E-ink support is built in, not bolted on - Kobo devices sync directly against Komga instead of Kobo's servers, with two-way read progress, on-the-fly KEPUB conversion via Kepubify, and optional proxying so official Kobo purchases still work; KOReader sync covers everything else. Multi-user management brings per-library access control, age restrictions, and label restrictions for family setups. Housekeeping tools detect duplicate files and duplicate pages, and a REST API feeds a healthy ecosystem of community scripts. Runs from a single Docker container with embedded SQLite.

Page views, referrers, browsers, and screen sizes - Ackee delivers the analytics developers actually check, from a deliberately minimal Node.js and MongoDB stack that skips both Matomo's weight and Google Analytics' cloud dependency. Its defining constraint is anonymization: no cookies, no unique user tracking, and a multi-step anonymization process that keeps visitors unidentifiable while the aggregate numbers stay useful. In its default anonymous mode Ackee collects no personally identifiable information at all, which means GDPR and CCPA compliance out of the box and no cookie consent banner on your sites. A detailed mode adds screen size, language, and per-visit referrers - still without cookies or fingerprinting. Integration mirrors the Google Analytics pattern: create a domain in settings, drop the generated ackee-tracker snippet into your pages, and data appears in a clean single-page dashboard. One instance tracks multiple domains, and custom events capture button clicks, signups, and conversions. The distinctive engineering choice is the fully documented GraphQL API: everything the dashboard shows comes from that API, so you can query active visitors, average duration, and view statistics programmatically, feed data in from apps and services beyond websites, or build an entirely custom interface on top. If you want bare-minimum analytics with a real API and zero privacy anxiety, this is the tool.

A deliberately simple ticketing system standing in for both Zendesk and Jira: Peppermint handles internal staff requests and external customer support alike. The stack is modern full-stack TypeScript: Next.js and React over Prisma and PostgreSQL, which makes it light to run and approachable for developers extending it. Ticket creation is straightforward - a markdown editor with file uploads, assignment, status tracking, and a logical workflow that new agents grasp without a manual. Mailbox integration converts email into tickets automatically: configure SMTP/IMAP per mailbox and incoming messages become trackable tickets. Each client accumulates an interaction history, giving agents context on every past request before replying. Two touches distinguish it from bare-bones ticketing: a built-in markdown notebook with todo lists for internal documentation and knowledge sharing, and OIDC authentication so agents sign in through your existing identity provider - Keycloak, Okta, Authentik, or Azure AD. Configurable webhooks and email notifications push ticket events to third-party services. The UI is responsive from mobile to 4K, and everything works fully offline in air-gapped environments. Docker-native and scalable via Kubernetes, with an active community of 3,000+ GitHub stargazers.

Subtitles are the one chore Sonarr and Radarr leave behind - Bazarr finishes the *arr media stack by automating them. It connects to both via their APIs and mirrors their libraries - it doesn't scan disk itself, it manages exactly what your *arr apps index. For every monitored episode and movie it checks existing internal and external subtitles against your language profiles, then hunts missing ones across dozens of providers - OpenSubtitles.com, Podnapisi, Addic7ed, Subscene, and many regional sources - covering 184 subtitle languages including forced/foreign-dialogue tracks. Matching is smarter than filename guessing: releases are compared by release group and source, some providers support exact file-hash matching, and every downloaded subtitle gets a percentage score. Set a minimum score per Sonarr/Radarr connection and Bazarr rejects weak matches; enable upgrades and it replaces previously downloaded subtitles when better ones surface. Out-of-sync files get fixed too - automatic subtitle synchronization realigns timing after download, triggered only below a configurable score threshold so good subs aren't touched. Per-show and per-movie language configuration, download history, manual on-demand search, and adaptive searching that throttles provider API calls round it out, all behind a clean Sonarr-style web UI written in Python. If your library serves multilingual viewers, this removes the last manual step.

Comment sections at the Washington Post, the Wall Street Journal, and newsrooms across 30 countries run on Coral (also known as Talk) - the platform built by journalists' technologists, started under the Mozilla Foundation and now stewarded by Vox Media as an Apache-2.0 project serving 23 languages. Its founding premise is that online comments are broken and moderation is the fix. Moderators get a full queue system - reported comments, system-held pending comments, and configurable pre-moderation - backed by AI toxicity scoring that warns commenters before posting and holds high-scoring comments for review, Akismet spam detection, banned and suspect word lists, and automatic repeat-offender handling that pre-moderates users whose rejection rate crosses a threshold. Readers get features designed for healthier conversation: journalist badges in threads, muting of annoying voices, notifications, instant new-comment alerts, and timeouts rather than just bans. For publishers the economics are the point - no ads, no trackers, no hidden pixels anywhere in the code, full ownership of audience data, and GDPR compliance beyond requirements. Integration is one embedded script; SSO connects existing registration, and a GraphQL API supports customization and extension.

Hold Alt/Option, click any string in your running app, and edit the translation in place: Tolgee is an open-source localization platform built the way developers wish translation worked - changes save straight to the platform with no hunting through JSON or PO files. In-context editing works even in production via the Tolgee Tools browser extension, which injects credentials without touching source code, so a client or colleague with zero coding skills can translate the product inside the product. The SDKs (React, Angular, Vue, Svelte, plus iOS and Android) extract context from each UI element and capture one-click screenshots, giving both human translators and machines the surrounding meaning that raw string files lose. Autonomous translation uses that context: new keys are instantly filled from translation memory or machine translation (DeepL, Google Translate, AWS Translate), with optional human review afterward - shipping no longer waits on a translation agency. A CLI handles import/export, a REST API covers automation, a Figma plugin bridges design, and an MCP server lets AI coding assistants search keys, create translations, and trigger machine translation without leaving the editor. Self-hosting this Crowdin/Phrase/Lokalise alternative keeps every string, screenshot, and API key on your infrastructure.

Configure each indexer once, not five times: Prowlarr is the indexer hub of the *arr stack, removing the most tedious duplication in a media automation setup. Instead of configuring the same torrent trackers and Usenet indexers separately in Sonarr, Radarr, Lidarr, Readarr, and Mylar3, you add each one once in Prowlarr and it syncs them to every connected app automatically, complete with category mappings; with Full Sync enabled, adding or removing an indexer propagates everywhere, and a tracker URL change is a one-place fix. Coverage is the deepest available: 500+ torrent trackers with definitions added constantly, 24 native Usenet indexers, generic Newznab and Torznab support for anything else, and custom Cardigann YML definitions with JSON and XML parsing for obscure sources. Built on the same .NET/React base as its siblings, it also earns a place as a search tool in its own right: unified manual search queries every indexer simultaneously at a category level, and you can push multiple releases straight to your download clients without touching Sonarr or Radarr. Health checks and status notifications flag failing indexers, per-indexer statistics track success rates and response times so you can cull unreliable sources, and per-indexer proxy support (SOCKS4/5, HTTP, FlareSolverr) handles trackers behind Cloudflare.

Bring your own API keys and work with OpenAI GPT models, Anthropic Claude, Google Gemini, Mistral, DeepSeek, Grok, Azure endpoints, and local models in one organized workspace: TypingMind is a unified chat frontend for large language models, replacing a browser tab per provider. Parallel chat sends the same prompt to multiple models and compares answers side by side, and models can be switched mid-conversation. A prompt library stores reusable, tagged prompts with variables, and the AI Agents system builds specialized assistants that bundle a base model, custom instructions, assigned plugins, and uploaded knowledge files for RAG. Plugins extend every connected model with web search, image generation (DALL-E, Stable Diffusion), Deep Research, URL reading via Firecrawl, and Zapier automation - plus MCP server integrations for Notion, Atlassian, and other external tools, and a JavaScript extension API for custom behavior. Chats store locally by default with optional sync. Self-hosting puts the interface on your own domain and, for teams, adds branding, member access limits, and shared prompt and agent libraries.

What Gmail did for email, Octobox does for GitHub notifications: an ephemeral, unmanageable stream becomes an inbox you can actually triage. GitHub marks notifications read the moment you glance at them and lets old ones vanish days later; heavy maintainers end up building elaborate Gmail filter systems just to cope. Octobox - a Ruby on Rails app over PostgreSQL - syncs your notifications into a persistent inbox with an explicit archived state: mark a thread done, and if the issue or PR sees new activity, it pops back automatically, so nothing silently falls through. Triage is keyboard-driven with Gmail-style shortcuts (j/k to navigate, e to archive, m to mute, s to star), and multi-select clears noisy repositories in bulk. Filtering is where it earns its keep: slice by repository, organization, type, action, state, reason, CI status, labels, author, assignee, or bot origin, combine prefix search filters, and pin favorite searches to the sidebar. The optional GitHub App enriches entries with live PR/CI status and labels so you can decide without clicking through. Auto-archive rules clear merged PRs and closed issues; muting and snoozing silence the rest. A REST API supports integrations. Self-hosting keeps your notification metadata - a map of everything you work on - on your own server.

A server full of scattered services becomes one clean application hub under Flame - a self-hosted startpage where every bit of configuration happens in built-in GUI editors, never a config file. Applications and bookmarks are created, edited, and organized into categories directly in the browser, and favorites pin to the homescreen for one-click access. The integrated search bar filters your apps and bookmarks locally as you type, and prefix shortcuts (like /g for Google) route queries to any of 11 web search providers or custom ones you define, making Flame a genuine browser homepage rather than a static link wall. Docker integration is the standout for homelabs: mount the Docker socket, add flame.type, flame.name, and flame.url labels to containers, and new services appear on the dashboard automatically - Kubernetes Ingress annotations work the same way. A weather widget shows temperature, cloud coverage, and animated conditions for your coordinates, and password authentication protects settings and items. Appearance is deeply customizable with 15 built-in color themes, a custom theme builder, and full custom CSS support. The stack is Node.js with SQLite behind a React frontend - light, fast, and inspired by the minimalist SUI design.

"A highly-opinionated midwestern self-hosted Discord music bot that doesn't suck," per its own README - Muse is built for servers the size of you, your friends, and your friends' friends. It exists because the big public music bots kept getting shut down or paywalled, and self-hosting yours means nobody can take it away. Written in TypeScript on discord.js, it joins voice channels and plays audio resolved from YouTube via yt-dlp, and given optional Spotify API credentials it auto-converts Spotify tracks, albums, artists, and entire playlists to playable equivalents. The playback details show real care: seeking within songs and videos, livestream support, local caching so repeated plays start instantly, volume normalization across tracks, and configurable volume controls including optional ducking that lowers music when people speak. SponsorBlock integration can skip non-music segments automatically. Users save favorite queries as reusable shortcuts, and one Muse instance serves multiple guilds simultaneously - one deployment for all your communities. Configuration is three environment variables (Discord token, YouTube API key, optional Spotify pair) and the personality is free: there is no vote-to-skip, because "this is anarchy, not a democracy," and the bot remains a loyal Green Bay Packers fan. MIT-licensed and easily extendable.

Agents with tool use, deep research, a document editor, an IMAP/SMTP email client with AI triage, notes, tasks, and a CalDAV-synced calendar - Odysseus bundles all of it into one open-source, self-hosted AI workspace. It runs local models through Ollama, vLLM, or llama.cpp and cloud APIs like OpenAI and OpenRouter, with a hardware-aware Cookbook that scans your machine and recommends quantized models that fit. Persistent memory uses ChromaDB with hybrid vector-plus-keyword retrieval, web search runs through a bundled SearXNG instance, and agents can use MCP servers, files, and shell access with safety controls, plus custom skills and scheduled agent tasks. A blind Compare mode runs side-by-side model duels with identities hidden and accumulates Elo-style ratings from your votes, so model selection is based on your actual workloads rather than leaderboard claims. Deep research mode - adapted from the Tongyi DeepResearch approach - reads sources through SearXNG and produces cited reports, while the email client tags, summarizes, sets reminders, and drafts replies locally rather than through a third-party mail AI. The writing-first document editor adds AI edits, Markdown and HTML support, and version history. The stack is Python 3.11 with FastAPI, SQLite for state, and a vanilla JS frontend, licensed AGPL-3.0 with zero telemetry. Because agents can read email and execute commands, keep authentication enabled and never expose it as a public unauthenticated service.

Live web search plus an LLM of your choice: Farfalle is an open-source, self-hosted answer engine in the Perplexity mold. Queries route through one of several search providers - self-hosted SearXNG for a fully independent stack, or Tavily, Serper, and Bing APIs - and the model composes a cited answer from the retrieved results. Model flexibility is the core design: run llama3, mistral, gemma, or phi3 locally through Ollama for zero per-query cost and full privacy, use cloud models like GPT-4o or Groq-hosted Llama 3 for speed, or route to any provider via LiteLLM. An Expert Search mode uses an agent that plans a multi-step search strategy and executes it for harder questions, and chat history keeps prior research sessions available. The stack is a Next.js and shadcn/ui frontend over a FastAPI backend with Redis rate limiting, shipped as a pre-built Docker image. A browser search-engine entry pointing at your instance makes it the default search from the address bar. Paired with SearXNG and Ollama, the whole pipeline runs with no external API at all.

Personal, minimalist, database-free bookmarking - Shaarli is a philosophy as much as an app. Everything lives in a single compressed datastore file inside data/: no MySQL, no PostgreSQL, backup by copying one directory. That write-once/read-many file is usually served straight from OS disk caches, which is why a decade-old Shaarli instance with tens of thousands of links still responds instantly. Designed deliberately single-user, it saves URL, title, unlimited-length description, and tags (with autocomplete, renaming, and merging), marks entries public or private, and automatically strips utm_source and fb tracking parameters from saved URLs. That description field is why the community uses Shaarli as far more than bookmarks: a microblog, read-it-later queue, code-snippet base, pastebin, and shared clipboard between machines. Sharing is one click via bookmarklet or Android apps; consumption is per-tag RSS/Atom feeds plus a daily digest feed; search is full-text with tag filtering. A REST API opens it to any client, a plugin and theme system extends the PHP core (Markdown rendering, thumbnails), and import/export uses browser-standard Netscape HTML - your data enters and leaves freely. LDAP login is supported, no telemetry is sent anywhere, and the UI degrades gracefully without JavaScript. The anti-cloud Delicious.

Receipts, contracts, and letters that arrive as pixels rather than text: Papermerge is a document management system built specifically for scanned documents. Upload a PDF, TIFF, JPEG, or PNG and OCR runs automatically, with a real-time status indicator beside the document title; under the hood it drives OCRmyPDF and Tesseract, supporting 130+ languages and producing a new document version with a selectable, searchable text layer you can download. Full-text search runs across everything, with multiple search-engine backends (Xapian by default). The interface deliberately mimics a modern desktop file browser: dual-panel commander, drag and drop, hierarchical folders, and colored tags on documents and folders. Page management fixes what scanners get wrong - delete blank pages, rotate, reorder, merge, extract, and move strayed pages between documents, powered by PikePDF. Document types act as categories, each with its own custom metadata fields, so invoices carry vendor and amount while contracts carry parties and dates. Versioning preserves every state of a document. Multi-user support includes groups, group ownership, permission management, and document/folder sharing between users and groups, and an OpenAPI-compliant REST API automates ingestion from scanners or scripts. Apache-licensed, ideal for long-term digital archives.

A single pane of glass for every service you run, arranged by drag-and-drop with no YAML or JSON files: Homarr is the modern dashboard for self-hosted infrastructure. Its grid system arranges apps, widgets, and bookmarks on desktop or touch, backed by an icon picker with over 11,000 icons. What separates Homarr from static launchers is 50+ live integrations: Plex, Jellyfin, and Emby media stats, the *arr suite (Sonarr, Radarr) with a unified release calendar, download clients, Pi-hole and AdGuard Home controls, Proxmox, Home Assistant, OPNsense, and Unifi monitoring. Widgets update in real time over WebSockets (tRPC and Redis under the hood), and a built-in search queries thousands of data points across connected services. Custom widgets extend the reach to any HTTP API without code: define endpoint, auth, and refresh interval in the management UI, then render responses as stat grids, tables, progress bars, status indicators, action buttons, or full custom JSX layouts - with an AI-prompt helper for generating templates. Multi-user support is first-class: credentials, OIDC, or LDAP sign-on, groups with granular permissions, and secrets encrypted with AES-256-CBC. A robust background-job system scales it from a Raspberry Pi homelab to deployments serving hundreds of users.