Draw a UI
Sketch a wireframe, get working code: Draw a UI turns hand-drawn layouts into web interfaces. It pairs the open-source tldraw canvas with an OpenAI vision model: you sketch a layout - boxes, labels, buttons, arrows, whatever communicates the idea - select the drawing, and click Make Real. The app snapshots your selection as a PNG, sends it to the vision API with instructions to return a single HTML file styled with Tailwind CSS, and renders the result in an iframe directly on the canvas next to your sketch. The loop is iterative: annotate the generated prototype or redraw parts of it, select both the sketch and the previous result, and generate again - the model receives the earlier HTML as context and produces an updated version. Built by Figma engineer Sawyer Hood as one of the first viral GPT-4 Vision demos and the basis for tldraw's "Make Real", it is a Next.js app that runs against your own OpenAI API key. Self-hosting matters here: the upstream demo ships without authentication, so a private deployment keeps your API key from being drained by strangers. MIT-licensed.
Multiwoven
When warehouse data never reaches the CRM, ad platform, or help desk that needs it, Multiwoven is the open-source reverse ETL platform that syncs it there on a schedule. Sources include Snowflake, BigQuery, Redshift, Databricks, PostgreSQL, ClickHouse, MariaDB, Oracle, Amazon S3, and SFTP. Models define what to move: write SQL with a 100-row preview, pick tables in a visual selector, or reuse existing dbt models. Destinations span Salesforce, HubSpot, Klaviyo, Mailchimp, Zendesk, Slack, Airtable, Google Sheets, Stripe, Facebook Custom Audiences, HTTP endpoints, and vector stores such as Pinecone, Qdrant, and Weaviate. Each sync maps model columns onto destination objects like Salesforce Account, Contact, or Opportunity, then runs as a full refresh or as an incremental sync that moves only new or changed rows, optionally tracked by a cursor field. Schedules can be manual, an interval from minutes to weeks, or a cron expression. A dashboard reports sync activity, each run records its status with successful and failed row counts, and failed runs can trigger status emails. Workspaces keep sources, models, syncs, and destinations separate for each team, while Admin, Member, and Viewer roles and email invitations control access. A connector framework lets teams add sources or destinations that are not yet supported. 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.
Mbin
Mbin is a federated community platform that mixes magazine-based link aggregation and threaded discussion with short-form microblogging, so one self-hosted instance works as both a Reddit-style forum and a Mastodon-style social feed. Members create magazines around topics, then post links, articles, photos, and videos into them, vote submissions up or down, boost the best ones to their followers, and argue the details in threaded comment trees. The microblog stream sits beside the magazines for quick status posts, and a combined view blends both content types into a single feed you can sort by hot, new, active, or most commented. Because the whole platform speaks ActivityPub, your magazines and users federate with Mastodon, Lemmy, Pixelfed, Pleroma, and PeerTube, and follows, replies, and votes travel across instance boundaries in both directions. Magazine owners appoint moderators, pin threads, and handle reports, while instance admins control registration, federation allow and block lists, and which remote servers the instance talks to. Readers shape their own view with subscriptions to magazines, people, and tags, plus personal blocks for users, magazines, and entire instances. A documented REST API with OAuth 2.0 access lets mobile clients, bots, and integrations read and publish content, and Service-type accounts are supported for automated posters. 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.
Medama
Track website traffic and conversion funnels with sub-second precision while guaranteeing visitors total anonymity with Medama, a privacy-first web analytics platform operating without cookies, IP storage, or consent banners. Site owners embed a sub-kilobyte tracking script into web headers to record page views, session duration, referrers, screen resolutions, browser user agents, and geographic countries anonymously. Built-in caching header mechanisms determine unique daily visitors without generating persistent hashes or storing cross-site identifiers that violate privacy regulations. The web dashboard visualizes hourly traffic trends with interactive area charts and provides multi-dimensional data tables breaking down top-performing pages, search engine referrers, operating systems, and campaign UTM parameters. Developers can instrument custom click events and page attributes using simple HTML data attributes to measure conversions, document downloads, button interactions, and user subscription states. Administrators manage multiple websites from a single interface, configure curated anti-spam blocklists using Firehol and Tor exit node feeds, and monitor server CPU, memory, and disk usage in real time. Full OpenAPI specification endpoints allow teams to export analytics data directly into internal business intelligence pipelines and custom executive dashboards. 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.
Biniou
Thirty generative AI tools behind one login on your own server: that is Biniou, a self-hosted studio for making images, text, music, video, and 3D objects without sending a prompt anywhere else. The chatbot module runs GGUF language models locally, a multimodal variant sees images, Whisper transcribes speech, and a translation module covers 200 languages. Image work spans text-to-image across Stable Diffusion, SDXL, Flux, PixArt, and Kandinsky model families, plus img2img, inpainting, outpainting, ControlNet, face swapping, upscaling with Real-ESRGAN, and face restoration with GFPGAN. Music arrives from MusicGen, MusicLDM, AudioGen, and Bark voice generation, video from Modelscope, AnimateDiff, and Stable Video Diffusion, and Shap-E turns text or images into 3D shapes. Modules hand outputs to each other, so a generated caption can become a prompt and a still can become a clip. Everything runs from 8GB of RAM with no dedicated GPU, works offline once models are downloaded, and a control panel inside the interface handles updates, restart, authentication, network access, and one-click sharing. Models are managed from the same panel, generation settings are saved as metadata in every file, and CUDA acceleration is there when hardware allows. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. GNU General Public License v3.0 licensed.
Proval
Proval is a self-hosted AI code review agent that reads every pull request on your GitLab, Forgejo, or GitHub server and posts a consolidated, severity-grouped review with inline comments, using a model you run yourself. When a pull request gets a meaningful push, a planning agent groups changed files into review units and dispatches specialist sub-agents in parallel, each exploring the codebase so cross-file issues surface. A writing agent merges their findings into one review, with comments on the affected lines, including multi-line ranges. Reply to any Proval comment and it reads the thread context and answers, and it works on issues too, greeting new ones and answering follow-ups on every message or only when mentioned. Choose per repository whether reviews run on first push only, every push, or not at all, skipping drafts until they are marked ready. Every review, reply, and issue comment is logged with its token usage, so each model's cost shows in the dashboard instead of hiding in an API bill. Your code is never stored; review context goes only to the LLM endpoint you configure, whether that is a cloud API or a local Ollama or llama.cpp server on your own network. Connect GitLab or Forgejo with an access token and webhooks, or GitHub through a GitHub App; Gitea and Codeberg use the same Forgejo-compatible path. 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.
Anysub
Need subtitles but not a cloud upload? Anysub transcribes audio and video with Whisper on your own server and hands back editable cues. Drop in a video or audio recording, pick a Whisper model from tiny to large, and the worker transcribes it with word-level timing. The studio then puts the video, the cue list, and a zoomable waveform timeline side by side, so every subtitle can be dragged, trimmed, split, merged, or retyped. Warnings flag cues that run too fast, too long, or overlap a neighbor, and autosave with undo keeps edits safe. Long uploads resume after a dropped connection, and the searchable library finds projects by name or by the words spoken inside them. Finished tracks export as SRT, WebVTT, plain text, or JSON with words and metadata. Administration pages manage model downloads, transcription defaults, upload limits, disk usage, and connected workers, and a first-run setup code protects the owner account. 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.
RSSBox
Foreign-language feeds are where good information usually goes to be ignored, and RSSBox fixes that by translating, filtering, and merging RSS sources, then republishing them as clean feeds your existing reader can subscribe to. Add a feed, pick a translation engine, and RSSBox republishes it with titles, full articles, or summaries translated, optionally showing both languages side by side. Each feed can use a different engine, from DeepL to any OpenAI-compatible model endpoint, and the admin shows translation status plus the tokens or characters each source has consumed. Update schedules are set per feed, so a fast-moving news source refreshes hourly while a quiet personal blog checks in once a day. Keyword filters quietly drop the recurring topics you never open, and AI filtering goes further by judging items against what you actually read. Tag several related sources and RSSBox merges them into a single unified feed, translations and filters applied, ready to follow as one stream. Full-text fetching pulls complete article bodies from feeds that only publish excerpts, so the translation covers the whole story rather than the teaser paragraph. Your reader app never changes: it simply subscribes to the translated feed URL like any other source. 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.
phpBB
No per-seat fees, no landlord, no shutdown notice: phpBB gives your community the complete bulletin board package on a server you own. Organize discussion into categories holding unlimited forums and subforums, each with its own rules, password, style, or redirect, and let members post in flat topics with quoting, BBCode formatting, file attachments, polls, smilies, and emoji. Members register through configurable anti-spam checks, subscribe to forums and topics, keep bookmarks, exchange private messages with folders and drafts, and get notified about replies and mentions in their notification center. Moderators work from their own control panel with approval queues, reported-post handling, warnings, topic tools, pruning, and bans by username, email, or IP range, while the famous permission system assigns forum, moderator, and admin roles per user or group and can trace exactly why any member can or cannot do something. Administrators run everything from the ACP: styles built on the responsive prosilver theme, extensions installed from the validated catalog, language packs for a global audience, search and cache backends, scheduled tasks, and database backups. Whether it is a support board, a hobbyist community, or a company knowledge commons, the conversation stays on infrastructure you own. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. GPL-2.0 licensed.
myCart
Sell from your own server with a store that arrives as a single file: myCart pairs a public storefront and an admin panel with checkout, and there is no plugin ecosystem to assemble first. You manage products with variants that track their own inventory, pricing, and SKUs, and you can sell physical goods, downloadable files, and license keys from the same catalog. Checkout accepts Stripe and PayPal for cards and wallets, SpectroCoin and Coinbase Commerce for cryptocurrency, and a built-in dummy provider that processes free products with no payment setup at all. Buyers can check out as guests or register for a customer cabinet that keeps their orders, re-downloadable files, and issued license keys in one place. The storefront ships in eight languages, English, Chinese, Korean, French, Spanish, German, Italian, and Belarusian, with the language switcher available in both the shop and the admin panel. You upload your own logo, favicon, and tagline from the admin panel and the storefront adopts them immediately with no rebuild. The whole shop runs on an embedded SQLite database with nothing else to operate, or on PostgreSQL when you need concurrent writers and managed backups. Built-in HTTPS support and hardened, separately keyed sessions protect the panel and the customer cabinet. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Bionic
Bionic gives a company its own private AI workspace where models can read documents, use approved tools, run code, and hand back finished files. Staff open a chat, pick a model, and ask for work; the agent searches datasets, calls connected systems including Gmail, Google Drive, Salesforce, ServiceNow, and Jira, and saves spreadsheets, reports, and PDFs as artifacts. Skills package reusable instructions for jobs like release readiness, incident briefings, policy gap reviews, and RFP responses, so a good workflow gets written once and reused by everyone. Datasets collect security policies, product docs, handbooks, and support archives for grounded answers. Administrators manage API keys, watch token usage and request charts, set rate limits, configure providers, and divide people into teams with their own assistants. SSO, role-based access, and an audit trail keep the rollout governed. Developers can issue model and assistant keys and turn any OpenAPI specification into a new integration. 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.
Grimoire
Grimoire captures, extracts, and indexes the content behind your bookmarks so you can search what pages actually say, not just their titles and URLs. The ingestion pipeline accepts links from the web UI, REST API, MCP server, browser bookmarklet, or bulk import, then fetches each page and extracts readable content using specialized parsers for GitHub repos, GitHub issues, StackOverflow threads, YouTube transcripts, PDFs, and standard web articles. Everything stores locally in SQLite with file-based content archives. Search operates in three modes: FTS5 keyword matching for exact terms, semantic embedding search for meaning-based retrieval using vector similarity, or a hybrid ranking mode combining both. Optional AI providers including OpenAI, Ollama, Anthropic, DeepSeek, and any OpenAI-compatible endpoint generate automatic tags, summaries, and embeddings without being required for core functionality. The interface built with React 18, Vite, TypeScript, Tailwind CSS, and Radix UI supports categories, nested tags, notes, archive and trash states, read-later flags, and multi-user isolated spaces. A single Bun-powered Hono process serves both the compiled frontend and the REST API on port 3210, requiring only one Docker container and a SQLite volume. Backup and restore export bookmarks, content, settings, and metadata as portable ZIP archives. Nearly 3,000 GitHub stars reflect growing adoption among developers and researchers. Running on a VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
TDuck
Describe the survey you need in one sentence, and TDuck drafts the questions for you. This open-source form platform replaces hosted tools like Typeform with a version you run on your own server, so responses never leave your hands. The visual designer offers over 27 field types, from text, dates, and ratings to file uploads, signatures, cascading selects, sub-forms, and matrix scales. Logic rules show or hide questions based on earlier answers, and imported question lists save time on long forms. Appearance settings cover header images, backgrounds, colors, and button text. Each form can be limited by device, account, IP, or WeChat identity, with an answer window you set. Finished forms publish as links or QR codes, and a recycle bin recovers deleted ones. Statistics show views, completion rate, weekly trends, submission maps, and per-question pie or bar charts that export as images. Responses can be edited, exported, printed, or zipped with attachments. Webhooks fire on new, changed, or deleted entries, a sync API exposes all data, and email, SMS, and WeChat notices keep owners informed. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
Nezha Monitoring
Keep every VPS, home-lab box, NAS, and router you run on one self-hosted dashboard, with a small agent on each machine reporting whether it is up and how hard it is working. Nezha tracks CPU, memory, swap, disk, load, traffic, process counts, and TCP and UDP connections on Linux, Windows, macOS, OpenWRT, and Synology machines, plus GPU and temperature where supported. Servers sort into groups and appear as cards, a list, or a world map, with live charts and, when the optional time-series database is on, 1, 7, and 30 days of history. HTTP GET, ICMP Ping, and TCPing monitors run from the servers you choose, chart latency and packet loss, and warn about HTTPS certificate changes and upcoming expiry. Alert rules watch offline servers, resource use, bandwidth, and cycle traffic limits, then message Telegram, Slack, Matrix, Feishu, DingTalk, or any webhook. A browser terminal and file manager handle quick fixes, while scheduled or alert-triggered tasks run backups, restarts, and scripts across many servers. DDNS updates Cloudflare, Tencent Cloud, or Hurricane Electric records when an agent's IP changes, and NAT traversal publishes a home router or NAS page on your own domain. Admin and user roles, scoped API tokens, and an optional MCP endpoint let people and AI tools list servers and run commands within their permissions. 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.
Dialoqbase
Retrieval-augmented chatbots on your own knowledge base - that is the whole mission of Dialoqbase, an open-source bot-building platform. Feed it content through a broad set of data loaders - web pages and full crawls, sitemaps, PDFs, DOCX, CSV, plain text, GitHub repositories, YouTube videos, and MP3/MP4 audio - and it handles the whole RAG pipeline in one self-contained app: chunking, embedding, vector storage, and LLM querying. The distinguishing architecture choice is PostgreSQL with pgvector for embedding storage and similarity search, which removes the separate vector-database dependency, and Redis-backed Bull queues for ingesting large documents without blocking the API. Model choice is wide open: OpenAI, Anthropic Claude, Google Gemini, Cohere, Fireworks, Hugging Face, local models via Ollama, and any OpenAI-compatible endpoint, with an equally broad list of embedding providers. Finished bots embed on any website with customizable styling or deploy to Telegram, Discord, and WhatsApp, and an API creates and manages bots programmatically. Multi-user support adds registration limits and per-user bot quotas. MIT-licensed and free for commercial use.
Live Helper Chat
Add a live support desk to any website with Live Helper Chat, where visitors talk to operators or bots and you own every transcript. Site owners paste a generated embed snippet into any page, and the start-chat fields are customizable. Chats are routed by department, language, location, or custom attributes, wait in a priority queue, and move between agents. Operators, departments, and simultaneous chats are unlimited. Agents watch a visitor type before the message is sent, read pages visited and time on site in the online visitors list, and send proactive invitations to people who linger. Voice, video, and screen sharing run through Jitsi or Agora, and a screenshot shows what the visitor sees. Bot designers chain triggers and actions in a visual builder, call ChatGPT, Gemini, Ollama, Rasa, or any REST API, and hand the conversation to a person when needed. Incoming email becomes tickets beside chats, and extensions connect Telegram, WhatsApp, Facebook Messenger, and Twilio SMS. Staff answer from the browser, a desktop app, a Chrome extension, or iOS and Android apps. Management tools include roles, groups, statistics, transcripts, and an archive, and extensions add two-factor login, LDAP, and single sign-on. It suits banks, healthcare teams, universities, and game studios that must keep chat data in house. 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.
LocalGPT
If a document is too sensitive for a hosted chatbot, LocalGPT lets you question it in plain English on your own server, with every answer citing the passages it used. Drop PDF, DOCX, HTML, Markdown, or plain-text files into a named index and scanned PDFs are read through OCR. Vector and full-text results are fused by rank, then a cross-encoder reranker keeps the best passages. A router decides per question between searching your documents and answering directly. Complex questions split into sub-questions that are searched separately, pooled, and answered in one pass. Progress cards stream the retrieval, reranking, and synthesis stages into the chat as they run. An optional verification pass appends a confidence percentage to the answer. A session-scoped semantic cache answers repeated questions faster. Per-chat settings toggle decomposition, pruning, verification, search type, model, and retrieval depth. Swap the answer model per session or per message, and choose the embedding model when you build an index. From a terminal, one command indexes a folder and another asks a question, and a REST API exposes every action the interface can take. A start script runs everything in Docker against Ollama on the host or in a container. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.
RESTai
Every AI project you build in RESTai becomes an endpoint that any app can call over plain HTTP. RESTai gives you a browser dashboard where each project gets its own model, system prompt, tools, and chat playground. Build knowledge-base assistants over uploaded files and synced web pages, ReAct-style agents that call MCP servers, or drag-and-drop block projects that need no model at all. Connect Ollama, OpenAI, Anthropic, Gemini, vLLM, Azure, Bedrock, or any OpenAI-compatible endpoint, and let teams share the ones you approve. Daily charts follow tokens, cost, and latency per project, with rate limits, guardrails, prompt version history, and evaluation runs for quality tracking. Teams, roles, two-factor sign-in, white-label branding, and an audit trail cover administration. A one-line chat widget embeds a project on any website, and connectors reach Telegram, Slack, and WhatsApp. One install can serve several branded assistants and bill each team separately. 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.