Every site you follow, folded into one calm river of headlines that you host yourself. selfoss is an open-source RSS reader and feed aggregator that gathers updates from websites, blogs, news outlets, and social platforms into a single reading stream on your own server. The newest, unread, and starred views keep triage fast: skim the river, star what deserves a slower read, and mark the rest done; tags with their own colours sort sources into topics, and full-text search finds the article you half-remember. Feeds refresh on a schedule you set through a cron job or timer, thumbnails and favicons are cached locally, and sharing buttons send finds to other services. Because it is plain PHP with SQLite by default, it runs almost anywhere, on the smallest hosting, with no database server to babysit; MySQL and PostgreSQL are there when the archive grows. Optional login protects the reader, public mode can share a stream, and the responsive layout reads as well on a phone as on a wide monitor. Your reading habits stay on hardware you control, and the archive keeps growing for exactly as long as you keep the server. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. GPL-3.0 licensed.

Close the laptop; the agents keep shipping. Cezar is an open-source control center that runs Claude Code, Codex, OpenCode, Copilot, and eight other coding agents in parallel on your own server, around the clock. Every task gets an isolated git worktree, so a dozen agents edit the same repository at once without colliding, and overflow waits in a queue. You watch each run live: the agent's text, its tool calls, token counts, and reported cost stream into one cockpit covering all your projects. The workspace dashboard separates what needs your answer, what is running, and what finished, and the usage view totals spend and tokens per project so a runaway task shows up in dollars, not surprises. Repeatable playbooks are plain Markdown skill files, chained into short YAML workflows that can mix a different agent per step, add checks and gates, and open pull requests; nothing merges on its own. Hand it a GitHub issue in one click, connect Jira or Linear to launch from your tracker, and schedule recurring maintenance with automations. Flip Autonomous and a run stops asking questions and just finishes; run the same task two or three times and keep the best diff. State lives in plain files, no database, and the responsive interface works from your phone. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Point OBS at it, and the same live stream comes out in whatever protocol each viewer needs. Xiu is an open-source live media server that accepts publishing over RTMP, RTSP, and WebRTC WHIP, then redistributes every stream as RTMP, RTSP, HTTP-FLV, HLS, or low-latency WebRTC WHEP playback. Cameras and encoders pushing H.264, H.265, or AAC over TCP or UDP become browser-friendly HLS and HTTP-FLV without a separate transcoding tier, and GOP caching gets late joiners to a keyframe fast. Static push and pull relaying chains servers together: one origin ingests, regional nodes pull, and audiences play from the node nearest them; RTMP clustering covers larger fan-outs. Every live stream can also be recorded straight to HLS files, giving you an instant archive of broadcasts, lessons, and town halls. Token authentication guards publishing and playback, HTTP callbacks report stream start and stop events to your backend, and an HTTP API answers stream information queries for your own dashboards and automation. Operators shape ports, protocols, relays, and recording in one configuration file or with command-line flags, running a single binary that stays out of the way until the audience arrives. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

A single binary watches everything you care about and only bothers you when something breaks. EaseProbe is an open-source health and status checker that probes websites over HTTP, TCP ports, ping, TLS certificates, shell commands, SSH hosts, and native clients for MySQL, PostgreSQL, Redis, MongoDB, Kafka, Memcache, and ZooKeeper. Each target gets an interval, timeout, retry checks, and data validation, from HTTP status and XPath evaluation to a Redis key matching an expected value; SSH host probes also watch CPU, memory, and disk so slow saturation gets caught alongside hard failures. When a target flips state, edge-triggered notifications fan out to Slack, Discord, Telegram, Microsoft Teams, email, AWS SNS, WeCom, DingTalk, Lark, SMS providers, log files, syslog, or your own shell script, with repeat alerts and quiet recovery behavior under your control. Availability history accumulates into daily, weekly, or monthly SLA reports delivered through those same channels, plus a live SLA page and JSON endpoint served by the probe itself. Everything it measures also exits as Prometheus metrics with per-target status, latency, and uptime, ready for Grafana dashboards and alerting rules alongside the rest of your stack. Configuration is one YAML file, persistence is one local data file, and the same binary runs on a tiny VPS watching an entire fleet. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Freelancers keep rebuilding the same handoff: developer builds the site, client breaks the layout trying to change a headline. Primo is an open-source CMS built for that handoff, where developers define reusable blocks and page types, and clients edit text, images, and links directly on the rendered page without touching the structure underneath. Multiple client sites live in one dashboard, organized into groups with a domain each, and publishing generates fast static HTML you can serve from Primo itself or export to another host. The whole site also exists as structured local files, so you or an AI coding agent can pull a site down, edit its blocks and fields in your normal tools, push changes back with revision checks that protect client edits, and publish when ready. Editors can add child pages and reorder the blocks a page type allows, while developers keep control of block code, field schemas, layouts, and roles. A marketplace of starters gives new projects a running start across restaurants, portfolios, practices, and shops. One small server binary with embedded storage runs the CMS and serves the published sites, which suits an agency hosting a dozen client sites on one machine. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Design the page on a canvas, publish it with one click, and stop paying a website platform for the privilege. Frappe Builder is an open-source visual website builder where you drag blocks onto a page, style them from a properties panel, preview desktop, tablet, and mobile side by side, and ship plain HTML and CSS that loads fast because no framework has to download first. Changes made in a tablet or mobile view apply only to that screen size, hover and focus states need no handwritten CSS, and color and font tokens carry light and dark values through the whole site. Because it runs on the Frappe Framework, pages can pull records straight from your database: a short data script fetches products or posts, a block repeats per record, and one design at a slug route becomes every product page or blog post. Headers, cards, and any block group save as reusable components with props, edited once and updated everywhere. Bob, the built-in AI assistant, drafts pages from a description using your existing colors, fonts, and components, asks before creating a form's storage, and files submissions where your team can see them. Publishing keeps version history, analytics count views and referrers without third-party scripts, images convert to WebP automatically, and pages can require sign-in. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Your team's code lives on someone else's server, under someone else's terms, until you decide it should not. CodeFever is a fully open-source, self-hosted Git hosting service that gives a small team unlimited repositories, branches and tags, merge requests, and branch protection on hardware as modest as one CPU core and a gigabyte of memory. Developers browse code with highlighted version comparisons, review changes through merge requests, and organize work with repository groups, member roles, and per-member permissions. Protected branches keep important history safe and make review the normal path to main, and webhooks connect pushes and merges to the other systems your team already runs. A super-admin backend manages every project and user on the instance, and the interface speaks Chinese and English natively. Installation is a single Docker command, updates preserve your data, and there are no repository counts, seat limits, or paid tiers waiting downstream. It suits a studio keeping client work private, a lab sharing models and notebooks, or a company that wants its source behind its own firewall. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

A rack of ordinary servers becomes your own cloud, with tenants ordering real infrastructure from a catalogue instead of filing tickets. Cozystack is an open-source, Kubernetes-based cloud platform that turns bare metal into a self-service system for spawning Kubernetes clusters, databases, virtual machines, load balancers, object storage, and HTTP caching services through a standard REST API. Hosting providers use it to sell managed services on hardware they already own, platform teams use it to give developers isolated environments inside quotas, and organizations leaving VMware run virtual machines and containers side by side with live migration and replicated storage. Tenants stay separated, GPU capacity can be shared between groups, and every service instance ships with preconfigured monitoring dashboards and alerts, gathered per tenant or into one hub. Because services are requested as Kubernetes resources, a billing system integrates by submitting a manifest, and platform packages are plain YAML delivered by FluxCD, so extending the catalogue means editing files your team already understands. Installation bootstraps immutable nodes over PXE or ISO, and the built-in web dashboard deploys applications and shows what the platform is doing without hiding the API underneath. The whole control plane runs on your hardware, auditable end to end. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Every self-hosted app asking for its own password is a liability you keep re-creating. Rauthy is a lightweight single sign-on identity provider that gives your whole homelab or organization one secure login through OpenID Connect and OAuth 2, with passkeys at the center instead of passwords. Accounts can be fully passwordless: an email address plus a FIDO2 passkey with user verification, or a password backed by a security key, with magic links and upstream login providers also supported. The admin UI manages users, clients, roles, groups, sessions, API keys, blacklists, and per-client login branding with custom logos and colors, and events stream to email, Matrix, or Slack when something deserves attention. Login endpoints defend themselves with brute-force and credential-stuffing detection, automatic IP blacklisting, rate limiting, and optional geolocation blocking. Beyond web apps, a PAM and NSS module extends the same identities to Linux workstations and SSH logins with ephemeral passwords or public keys, and the OAuth device flow covers headless and IoT devices. It runs as a single efficient service on embedded Hiqlite or Postgres, clusters for high availability, needs only tens of megabytes of memory, and has passed an independent security audit. Dynamic client registration, SCIM provisioning, and DPoP tokens cover modern decentralized login flows. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Training runs stop being terminal scrollback and become a board you can read at a glance. SwanLab is an open-source experiment tracking and visualization platform for AI model training that records metrics, hyperparameters, logs, and hardware signals from your training code into a modern web dashboard you can self-host. A few lines of Python start an experiment, and the dashboard follows loss curves, learning rates, and custom scalars live, with line chart configuration, smoothing, grouping, and detailed hover inspection for comparing runs. Experiments organize into projects and groups with tags, pinning, filtering, and a table view that sorts and aligns runs against a baseline with difference percentages. Media logging goes far beyond scalars: images, audio, video, text with Markdown rendering, 3D point clouds, molecular structures, and custom ECharts visualizations including PR curves, ROC curves, and confusion matrices. Hardware monitoring records GPU, CPU, memory, disk, and network behavior from NVIDIA, AMD, Ascend, and other accelerators alongside the metrics they produced. Integrations cover more than fifty frameworks, from PyTorch Lightning and Hugging Face Transformers to LLaMA Factory, veRL, and XGBoost, and existing histories migrate in through sync tools for MLflow, TensorBoard, and Weights and Biases. Teams collaborate with organizations, roles, project sharing, webhooks, and notification plugins. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

Give an AI agent a price list and it drafts the quote; hand it an invoice and it checks the numbers against purchase and receiving records before anyone pays. Atom is a self-hosted AI agent platform for business automation that wraps every agent action in governance: permissions, ordered policies authored in a canvas editor, human approval checkpoints, and durable audit records. Work starts from your files and instructions, with demonstrated workflows that map unfamiliar spreadsheet columns, calculate against source records, and produce cited Word or PowerPoint documents you can inspect line by line. Corrections become reusable teaching: saved field mappings, commercial terms, and contextual lessons carry into new work, and each job keeps the accepted snapshot it started with. Questions, approvals, and waits survive restarts, completed operations are retained, and uncertain effects must be reconciled before anything retries, so long-running business processes do not silently double-act. Results keep provenance back to the source records and operations that produced them, and unresolved requirements block completion instead of being invented away. You bring the model: configure a hosted provider or a local OpenAI-compatible endpoint, route by capability and cost, and run everything in a single-tenant workspace with user, role, and workspace boundaries. A starter sample runs with no model or cloud key at all. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. AGPL-3.0 licensed.

Where AI agents and people share one real-time workspace, an agent blocked on your decision never waits unnoticed until morning, and that workspace is AgentRQ, a human-in-the-loop task manager for agent fleets. In that workspace agents pull assigned tasks over MCP, report progress as they work, ask questions in the open, and pause visibly instead of going dark. Cards travel from Not Started to Ongoing to Blocked to Done on a list or Kanban board, and a Blocked card returns to work the second you answer from any device. Sensitive tool calls wait for an Allow click; flip YOLO on a task and they proceed automatically, still recorded call by call in a searchable trajectory. A timeline under every task totals how long the agent worked, waited, or stayed blocked. Scheduling spawns tasks on a cadence, completed tasks fire named events, and events hand new tasks to subscribed workspaces, chaining a release pipeline without glue code. Each project lives in an isolated workspace with its own agent, memory, skills, and Slack channel, and workspace memory carries yesterday's lessons into today's session. Voice-dictated tasks transcribe on device, 55 ACP-compatible agents connect through one gateway, and agents running on your own machines expose their terminals and resource use remotely. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. AGPL-3.0 licensed.

Versus Incident is a self-hosted AI SRE agent and incident manager that learns what normal looks like in your logs, metrics, traces, and Kubernetes clusters, then opens a triaged incident the moment something genuinely new appears. Detection earns trust gradually. Training mode only observes and catalogs patterns; shadow mode records what it would have alerted so you can judge its instincts; detect mode creates real incidents with an AI-written summary, a severity, and suggested next steps attached. Existing alerting does not have to move: Alertmanager, Grafana, Sentry, CloudWatch, Fluent Bit, and custom scripts post to the same webhook endpoint, and every incident, however it arrived, is rendered with your Go templates. Delivery fans out to Slack, Microsoft Teams, Telegram, Viber, email, and Lark. Silence has a deadline, because unacknowledged incidents escalate to on-call through PagerDuty, AWS Incident Manager, Opsgenie, incident.io, or ServiceNow. The Kubernetes dashboard gathers cluster health and capacity, workloads, resource relationships, recent changes, Helm releases, and GitOps status, and a DevOps chat agent in the web UI answers questions against the connected systems while responders work. The admin dashboard reviews learned patterns, shadow events, and raised incidents, so the agent's judgment stays inspectable. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

WhatsApp becomes an HTTP service your own systems can call. WPPConnect Server wraps WhatsApp Web sessions in a REST API for integrators who need to send and receive messages, manage contacts and groups, and react to events without building browser automation themselves. Each session pairs by QR code and gets its own token, and multiple sessions run side by side for different numbers, teams, or customers. Endpoints cover text and media messaging, contact and group operations, product catalog actions, and session lifecycle, all documented in a built-in Swagger UI and a Postman collection. Webhooks push incoming messages and status changes to your backend, with destination allowlists keeping callbacks pointed only where you configured. Chatwoot connects through a dedicated authenticated webhook, so support conversations can flow between the help desk and WhatsApp in both directions. The Docker image bundles WPPConnect Manager, a web workspace for session administration, QR pairing, and reading and answering chats, protected by the server secret key at the top level and per-session tokens below it. Credentials, browser data, media, and logs persist in named volumes, surviving container restarts without re-pairing. Support desks automate ticket updates, shops send order notifications, and clinics confirm appointments through numbers they already publish. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

One page, every place you exist online, hosted on a domain you own. LittleLink Server is a self-hosted link-in-bio page that publishes your profile, avatar, bio, and social buttons from a single Docker container, with no account fees and no third-party branding on your links. More than sixty brand buttons cover the major social, streaming, music, and creator platforms, and custom buttons add any destination with your own text, colors, icons, and ordering. Themes restyle the page in a few environment variables, and because configuration is read at request time, edits take effect on a container restart with no rebuild. Page metadata, Open Graph titles and images, favicons, and language are all configurable, so the link you paste into a bio unfurls properly in chats and feeds. The server generates its sitemap and robots rules at request time, with controls for index status and extra crawler rules, including blocking specific AI crawlers by name. Analytics support connects the page to your tracking provider, and a health check endpoint keeps uptime monitors and container orchestrators informed. Creators, streamers, developers, and small businesses use it as the stable home base behind every profile bio: one short URL that always points at the current set of destinations. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Files belong to the project, not to whoever uploaded them: ownCloud Infinite Scale is a self-hosted file sharing and collaboration platform where teams work in Spaces, data rooms that stay intact when people join, leave, or change roles. Inside a Space, members browse, upload, version, and share files from the web interface or synced clients, with full control over links, permissions, and expiry. Documents open for real-time editing through the web office integration of your choice, favorites and recent files keep daily work one click away, and full-text search with tags finds content across every Space you can access. Personal files live alongside team Spaces, a recycle bin catches mistakes, and administrators manage users, groups, roles, and storage from the same deployment. One service runs the whole platform for a home server, and the same system scales out for schools, companies, and public institutions that must keep data on their own infrastructure. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache-2.0 licensed.

One server, three hundred applications, zero Kubernetes: Websoft9 is a self-hosted platform that turns a single VPS into your private app store, with a web console for finding, installing, and running open source software. Browse the catalog, click install, and applications arrive as containers with their databases and settings wired up, ready behind your domain. Bind domains and issue free Let's Encrypt certificates through the built-in proxy, schedule local backups for apps and databases, and open a browser terminal or file manager when you need hands-on control. The overview shows what is running and what it uses, My Apps keeps every service one click from start, stop, or rebuild, and updates roll out from the same console. Small teams use it to stand up WordPress, GitLab, Mattermost, and internal tools on one machine instead of renting a separate hosted plan for each, and everything stays on hardware you own. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. LGPL-3.0 licensed.

Every Tuesday, a fresh fifty-song playlist lands in your own music server, built from what you actually listened to: Explo is a self-hosted discovery engine that turns ListenBrainz recommendations into playlists inside Navidrome, Jellyfin, Plex, and other home music systems. It fetches Weekly Exploration, Weekly Jams, and Daily Jams, then goes and gets the tracks, requesting songs from YouTube or Soulseek or whole albums through Lidarr, tagging downloads with proper metadata on the way in. Custom playlists import from Spotify, Apple Music, and ListenBrainz, so a list a friend made elsewhere becomes local files and a native playlist in your library. The web interface handles setup, schedules, and import history; last week's playlist can be kept for later listening instead of overwritten, and notifications report each run. For people who left streaming behind but miss algorithmic discovery, this closes the loop between a scrobble history and a self-hosted collection that keeps growing on its own. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.