MulmoTerminal is a browser cockpit for running many AI coding agents at once, laying every session out in a color-coded grid so you can see at a glance which agents are working, which are finished, and which are sitting idle waiting on you. Each cell is a real terminal running Claude Code, Codex, Muse, GitHub Copilot CLI, Cursor CLI, Grok, or Antigravity, and its border turns amber with a chime the moment a permission prompt or question blocks progress. Zoom into one session and the cockpit roster keeps the rest in view as one-line summaries with your last prompt, the agent's latest reply, and each branch's pull request phase. Agents hand you more than text: a Canvas panel renders documents, forms, charts, and generated images they produce, while per-session context, token, and cost readouts keep spending legible. Isolate parallel work in one-click git worktrees, review the diff, then commit, push, and open the pull request without leaving the grid. Sessions persist through tmux across restarts, Web Push pings your phone when a task finishes or needs input, and per-directory themes and name badges keep a wall of agents readable. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

TG Vault turns Telegram into the front door of your private archive, pulling files out of channels and groups into storage you control and managing them from a proper web file manager. Batch-download channel and group media by date range or tag, subscribe to sources so new posts sync automatically, and upload, search, organize into folders, and preview media from the browser. Every transfer runs as a managed task with live progress, failure details, and pause, resume, and retry controls, whether it was started from the web console or the Telegram bot. Point the vault at local disk, OneDrive, Google Drive, Amazon S3, Aliyun OSS, WebDAV, or an OpenList instance, and let the optional qBittorrent extension forward finished downloads into the same storage, cleaning up sources or keeping them seeding. Access stays locked down with an administrator password, a bot user allowlist, and TOTP two-factor authentication, while one install script deploys and upgrades the whole Compose stack. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Torollo is a hands-on backend engineering lab where the architecture you draw becomes real infrastructure, so a load balancer on the canvas is an actual Nginx container and a firewall rule you sketch is enforced by iptables. Place Ubuntu servers, PostgreSQL, MongoDB, Redis, and RabbitMQ nodes, wire them into VPCs and subnets, then watch the traffic simulator send requests through the topology you built. Guided roadmaps walk you from a single web server to a resilient three-tier application, a cache-aside Redis design, and a worker queue, grading every step against the live system instead of your intentions. Validators probe container status, SQL schemas, Redis keys, network reachability, and HTTP responses, and a failed check explains exactly what was expected versus what your containers actually did. Built-in explorers let you run SQL against Postgres, query MongoDB collections, and issue Redis commands, while browser terminals give you a shell inside any node. Roadmaps are plain JSON files you can import or author yourself, progress is tracked per step with hints and solutions, and the whole catalogue is available in English and French. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Slashdocs gives AI agents and the people who supervise them one shared home for finished work, replacing reports pasted into chat threads with reviewable HTML documents that live at a permanent address. An agent publishes a single self-contained HTML file, an architecture review, incident postmortem, product spec, data analysis, or QA report, and teammates open it in the browser to read the rendered page rather than raw markup. Reviewers anchor comments to exact words, resolve or reopen threads, and follow links straight to a heading or discussion, while Slack notifications announce new activity with personal mute controls for quieter channels. Every publish creates an immutable version guarded by optimistic concurrency, so an agent can read the feedback and push the next revision without overwriting anyone's copy. Agents connect through a remote MCP endpoint with browser OAuth or service-account keys, and people sign in through OpenID Connect with providers such as Google, Okta, Entra ID, or Keycloak. The companion CLI creates, pulls, pushes, and comments on documents with compact output or JSON for scripts, and folders plus sharing controls keep a growing library organized. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Every cloud account grows the same way: a test instance nobody shut down, a storage bucket from a migration two years ago, a database whose owner left the company. Steward walks all four of your clouds, AWS, Alibaba Cloud, Google Cloud and Microsoft Azure, and writes down what is actually running in one inventory you can search instead of four consoles you half remember. The point is not the list, it is what the list connects to: open any resource and see its relationships, which services depend on it, what shares its network, what breaks if it disappears. Cleanup stops being a leap of faith, because every candidate deletion is reviewed against that dependency map first, showing the impact before execution and following the result of each action afterwards. Scheduled scans keep the picture honest without you remembering to look, reporting exactly which resources each scan added, removed or modified, and pinging Slack, Feishu, DingTalk or WeCom when a scan fails. Connect accounts, including through OIDC, run the first sweep, and the orphaned, idle and forgotten start standing out from the infrastructure you actually need. It runs as a single executable with the web console and database built in, under systemd on your server, behind your HTTPS proxy with token sign-in. 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.

Running local models usually means choosing between a polished app that hides the engine and a terminal full of flags you half remember. LlamaForge refuses the trade: it installs and drives the official llama.cpp release itself, then puts a browser panel over all of it, every one of the 200-plus llama-server flags your build supports, grouped, searchable and editable per model. Before you load anything, each model is drawn into your GPUs at its real footprint, so the out-of-memory surprise happens on screen instead of in a crashed server; before you download anything, every quant on Hugging Face gets a fit rating for your VRAM, FITS, TIGHT or CPU OFFLOAD. Auto-Forge then measures placements, KV cache types and speculative drafters on your own hardware and keeps the fastest configuration, checking that a drafter still produces identical tokens before applying it with one click, and undoing it with one click. Keep several models loaded at once and a planner places each on the GPUs that fit it; when a game launches, Game Mode unloads your models and brings them back when you quit. Chat in llama.cpp's own interface, join from your phone by QR code, duel two models blind in the Arena under an Elo board, and let Embers, small local agents, keep a wiki and write scheduled briefs counting only facts they can quote verbatim. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Planning a weekend trip, a household move or a club event in a group chat means the packing list is forty messages up, the poll about dates expired, and nobody remembers who volunteered to book the van. Plainspace gives that chaos one calm page: create a Space, share a single link, and everyone lands on the same shared board of tasks, notes, checklists, polls and availability, in any browser, with no app to install and no account to create. One person adds the campsite options as a poll, another claims the cooking tasks, a third drops the address and gate code into a note, and the whole group watches it come together with live updates as people type, vote and check things off. Optional push reminders nudge the person whose task is due before the plan quietly stalls. It is deliberately not a project management suite: no sprints, no story points, no permission matrices, just the handful of primitives real life actually coordinates with, presented touch-first so it works as well on a phone at the supermarket as on a laptop. Run it for your family, your team or your club on the hosted version, or self-host the same thing with Docker Compose next to the rest of your services. 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 dedicated backup server is the right way to keep copies on separate hardware, and a strangely expensive one to leave humming in the corner for a job that runs twenty minutes a night. Joulenap lets that machine sleep: at the scheduled hour it sends a Wake-on-LAN packet to your Proxmox Backup Server, waits until the API answers, runs the backup, applies retention, runs garbage collection and verification if you asked for them, then powers the box back down or suspends it for network cards that only wake from sleep. You describe your setup as routes: which hosts back up to which backup server, on what schedule, keeping how many copies, with guests as everything on a host or an explicit list. Fan several hosts into one server nightly, or copy one backup server to another across town for an off-site second copy with its own retention. A second route kind starts nothing at all: your Proxmox hosts keep their own schedules while Joulenap simply wakes the box, watches their tasks, and switches it off when they go quiet. Watch live task output, stop a run mid-flight if you must, and get results through Apprise, Telegram, ntfy, Discord or email per route. Status widgets feed Homepage, Homarr, Dashy and Glance, and a Prometheus endpoint lets Grafana alert you when a guest stops being backed up. 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.

Game night keeps dying in the group chat because nobody can be in the same room, and the one willing game master is burned out by October. DiceFrame puts an AI game master on your own server instead: open the web UI, describe what your character does in plain language, and the engine adjudicates the action, rolls the dice check once, applies the result to your character sheet, and narrates what happens next. Run a solo campaign at midnight or invite friends through a link and play the same shared world from different cities, with waiting states, away players and live updates keeping the table in sync. Rules are layered rather than decorative; D&D 5e-inspired d20 play with advantage and disadvantage, CoC 7e-inspired d100 with bonus and penalty dice, custom d20 systems and freeform no-dice narrative all track hit points, gold, items, experience, death and revival properly. Build worlds, lorebooks and adventures in the content library, import scripts, PDFs and Foundry or Fantasy Grounds modules, or let AI assistance draft the world, rules and characters for you. Character cards are reusable across games, long campaigns are held together by session summaries and optional memory recall, and the story can be read aloud with text-to-speech through system voices or your own endpoints. Plugins and chat bots extend the table, all through your chosen OpenAI-compatible model endpoint. 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.

Reading a new manhwa chapter on release day usually means waiting for a scanlation group, or pasting panels into three tools yourself. XianScan collapses that whole relay into one server: point its browser extension at a chapter on a reader site and it captures every page, or drop raw pages into the studio and let the pipeline run. Speech bubbles, captions and sound effects are detected automatically, the text inside them is read by OCR supporting ten languages in vertical and horizontal layouts, and tall webtoon strips are resliced along natural panel gutters so no bubble is ever cut in half. Translation keeps a sliding memory of previous pages so speakers, pronouns and story topics stay consistent, while terminology glossaries lock character names and cultivation terms across an entire series. The original lettering is then removed by neural inpainting that rebuilds the artwork underneath, and the translation is typeset back into the bubbles with computed font sizes, line breaks and per-script fonts, including right-to-left scripts and accent fonts for attack names. Inspect OCR boxes and confidence scores, fix text before saving, then read in the web studio or stream chapters to the Mihon app on your phone. Local models through Ollama or LM Studio keep everything offline, or plug in Gemini, OpenAI, Groq or OpenRouter keys when you prefer the cloud. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Most traders know their balance and almost nothing else, because the patterns that actually cost money hide across hundreds of individual fills. Trading Journal AI rebuilds your round-trip trades from a broker CSV, grouping partial fills, scale-ins and shorts automatically, then tracks win rate, profit factor, expectancy, drawdown, MFE, MAE and exit efficiency for every account. Thinkorswim and Interactive Brokers import directly, seven more brokers including Tradovate, NinjaTrader, Webull and tastytrade are in beta, and a generic template covers any other broker's export. The Leak Finder ranks groups of trades that share something you knew at entry, an entry window, a weekday, a size habit, an emotion from your diary, and shows which of them quietly drain the account, with confidence scored against the rest of your history. Exit Lab replays your completed stock trades on one-minute prices and compares your actual exits against 1R and 2R targets, breakeven rules, trailing stops and VWAP or EMA exits, in R and in dollars. If you add an Anthropic key, Claude reads your trading diary and grades each day on process rather than outcome; without one, every number still works, because the statistics are plain code and never AI. Four interface themes, one SQLite file, no account and no telemetry complete the picture. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

Years of bank statements, pay stubs and receipts sit scattered across dozens of websites, each demanding a login and offering one PDF at a time. PaperPull fetches that whole history for you: sign in yourself in a real browser window, answer the two-factor prompt, and it downloads statements, receipts, pay statements, W-2s and tax forms from 73 supported providers including Chase, Amex, ADP, Amazon, Walmart, Vanguard and Verizon. The tool only ever downloads; it never pays a bill, moves money or changes a setting, and that read-only promise is enforced by construction rather than by policy. Every archive is checked for freshness and for holes in the middle, so a series that looks current but quietly skipped three months is flagged by name, with per-account cadence learned from your own documents. Receipts become one long spreadsheet of every item you have ever bought across every store, with prices and dates, while transactions are read out of statement PDFs into workbook rows that are verified against each statement's printed beginning and ending balances. New documents can flow straight into a Paperless-ngx consume folder, everything runs under a password you choose on your own network, and no cloud account is involved anywhere. 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.

Your router calls it "Unknown device" at 192.168.1.23; Vedetta calls it the kitchen light and shows its work. Vedetta is a Home Assistant app that scans your local network, identifies what each device actually is, and explains every identification. One scan lists every device it can find with presence and latency history, in tile or list view, light or dark, on a computer or a phone. Behind each name sits the recognition engine: it weighs many weak clues together, counts a repeated fact once, groups devices by family, and shows certainty honestly, keeping anything doubtful in Other rather than inventing a label. Tap the info marker beside a name, brand, or type and a pop-up lists every clue, its weight, and the hypotheses that were rejected. Names or rooms you set by hand are never overwritten by later scans, and sleeping phones are remembered instead of flickering in and out. Choose from three search levels depending on how hard you want it to probe, let the nightly deep search refresh the picture, and use the network doctor to spot duplicate addresses, repeated restarts, devices that drop together, or a slowing gateway. It can also publish what you choose to Home Assistant and MQTT, display devices from other networks your Home Assistant already sees, and mask exports before they leave the 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.

Where is ChatGPT Plus cheapest this week, and which seller actually honors the warranty? Price Comparison (Zhitu Shijia) is a self-hosted platform that gathers AI subscription and account offers from many card shops into one catalog you can search, compare, and question. Tell the built-in guide your budget and need, for example a Plus top-up under a monthly cap with warranty cover, and it filters offers, explains recharge channels and purchase conditions, and narrows the list through follow-up questions. Browse by platform, delivery method, term, and warranty to compare prices and stock for ChatGPT and Codex, Claude, Gemini, Grok, and Cursor offers side by side, including finished accounts, shared accounts, and redeem codes. Price history shows how an offer moved before you buy, and official regional prices give a reference point when a quote looks suspiciously cheap or expensive. Email accounts, SMS verification, and residential IP offers sit in the same catalog, so the extras a subscription needs can be priced in the same pass. Install the comparison Skill or connect the MCP service and your own AI assistant can search products, compare merchant quotes, and pull price history with source links and collection times attached. In the admin console you add and review shops, run collection, and supervise the pipeline that identifies, classifies, and publishes products into the catalog. 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.

Your coding agents are scattered across a laptop, a server, and whatever machine you left running at home, and Agenvo is the self-hosted relay that lets one AI assistant reach all of them through a single MCP endpoint. Ask ChatGPT on your phone to check on Codex running on your server, or have an agent in Herdr on your laptop fix a failing test, from wherever you are. Connectors for Herdr, Codex app-server, Paseo, and Lody dial out to the relay over WSS, so the computers running agents need no inbound ports, no VPN, and no port forwarding. Your assistant sees the agents running in Herdr without restarting them, resumes Codex threads later, and starts new work where it fits: a Codex thread on the server, or Codex, Claude, or Devin in a Herdr pane. When an agent finishes a turn, waits for input, or exits, the relay pushes a change event back, so a watching assistant or ChatGPT dot reads the output and takes the next step. Closing the chat or losing the network does not stop the agents; connectors reconnect on their own, and your assistant picks up from each agent's current state. Deploy the relay on a single VPS with Docker or on Cloudflare, authorize MCP clients with OAuth over Streamable HTTP, and discover targets and method schemas with search before calling execute. 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.

Patching the same Android apps by hand every time a patch repository updates is a chore with your name on it, and AutoMorph is the self-hosted patcher that takes the chore over and rebuilds your apps on schedule. It monitors patch repositories such as ReVanced, Morphe, and community repos for updates, downloads base APKs automatically with Google Play first and APKCombo, Uptodown, and APKPure as fallbacks, and applies your chosen patches without you touching a command line. The Add Package wizard walks you from a package name through suggested repositories, patch selection with per-patch options, and aliases that map same-role patches across repositories so two repos stop fighting over the same job. Builds run once a day at a time you choose, or start manually from the web portal while live build logs stream to your browser. The Patcher Control Center shows every tracked app's version, patch status, and download links, and email alerts carry direct download links whenever a patch succeeds. Distribution goes further: one-tap Obtainium import by QR code or deep link, plus an F-Droid compatible feed for Obtainium and Neo Store, deliver updates straight to your phone, and a custom keystore keeps them seamless. A low-spec mode trims heap and thread limits for small servers, retention settings clean up old builds, and Pangolin, Tailscale Funnel, or Cloudflare Tunnels expose the portal over HTTPS. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. MIT licensed.

If your app asks an LLM the same routing, triage, or moderation question thousands of times a day, stuntd is the local proxy that learns those answers and stops paying for them. It sits in front of OpenAI or Anthropic as a drop-in replacement for the provider base URL, recording typed requests while relaying everything else byte for byte, and distils each decision site into a small head on a frozen Laya encoder. Once trained, the head answers in about 20 milliseconds with no tokens and no round trip; when its calibrated confidence falls below the site threshold, or the novelty gate flags the input, the question goes back to the provider instead of being guessed at. It speaks four protocols on one daemon: Jev System One with choice, score, and noul questions, the OpenAI Decisions API, Chat Completions, and the Anthropic Messages API, and every response carries an X-Stuntd header showing who answered. Run it keyless and the base Laya checkpoint answers zero-shot locally, put the paid Jev API in front to teach from real traffic, or import labelled rows and train directly. A few thousand rows close the gap: in the banking demo, the risk question climbs from 30.0 to 72.5 percent. A shipped Claude Code PreToolUse hook answers "may this command run" in about 50 milliseconds, without sending every command to a provider. 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.

"Run the migration and call me if anything fails" is an instruction your agent can finally obey, because Agentring is the self-hosted bridge that lets AI agents ring your phone and talk it through with you. When an agent needs a decision, hits a wall, or finishes a long job, your phone rings even locked and in Doze; a realtime voice AI explains why it is calling, answers your questions, and hands the transcript back so the agent carries on with your decision. You can also call your agents back from the Android app and ask what CI said while driving; the voice AI can consult your other agents mid-call, and a slow answer arrives when ready, or as a message if you have hung up. A missed call turns into a message you can answer from the notification. Anything that speaks MCP can reach you, including Claude Code, Codex, opencode, OpenClaw, and Hermes, and scripts or n8n reach you through the REST API and CLI. Pick your voice: Gemini Live by default, OpenAI Realtime, Grok Voice, Phonic, or your own local speech server with no key at all. The relay and LiveKit run in one Docker Compose behind Caddy, the daemon sits beside your agents and connects outbound only, and FCM push wakes the locked phone. No phone numbers, no Twilio, no per-minute fees. 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.