Stateless MCP servers: what actually changed
MCP revision 2026-07-28 removed sessions and the initialize handshake. What that means for your reverse proxy, health checks, timeouts and auth.
Run the software yourself, on your own server. Docker and Compose setups for the apps people most often self-host, with storage, TLS and backups covered.
Filtering by topic #self-hosting · clear
MCP revision 2026-07-28 removed sessions and the initialize handshake. What that means for your reverse proxy, health checks, timeouts and auth.
Which office server belongs next to Nextcloud: vendor RAM floors, the CODE connection limit, docx fidelity, and what really fits a 2 to 4 GB VPS.
Deploy Octop on a VPS with Docker Compose pinned to a tag: per-user isolation, an OpenAI-compatible model backend, TLS, and why to skip the curl installer.
Paritok compresses the file reads and tool output your coding agent sends. The project claims 74% fewer tokens. Here is the mechanism and the break-even math.
Vaultwarden encrypts every vault item in the client, so the server holds no plaintext. The real risks are your admin token and your backup file.
Syncthing is peer to peer with no server copy. Nextcloud is a server with accounts and links. Which one fits your VPS, and why neither is a backup.
Advertise a private network to your tailnet from a VPS: route approval, IP forwarding that survives a reboot, and the --accept-routes flag Linux needs.
Run the NetBird mesh VPN on one VPS: DNS and TLS setup, the pinned quickstart script, setup keys for unattended peers, and how it compares to Headscale.
PUID and PGID are not Docker settings. They are an entrypoint convention in linuxserver.io images. Why bind mount files land as 911:911, and how to fix it.
Run KiroCrew as a pinned container on your own VPS so memory and schedules survive reboots. Docker, systemd, SSH access, backups and rollback.
Copy a skill into eight repos and the copies drift. Treat skills like dependencies: one shared repo, and a version each project pins and reviews.
Run OpenTag on a VPS so Slack and GitHub mentions reach your coding agent: TLS ingress, webhook signature checks, token scopes, and safe defaults.
Planka, Vikunja, Focalboard, Wekan and Kanboard compared on RAM floor, database, SSO, Trello import and which of them are still maintained.
Build a self-hosted stock research agent on a VPS: a market data feed, a DuckDB store, a systemd timer that fires at market close, and an LLM screen.
Draco, Hound and self-hosted Firecrawl compared on RAM, headless browser needs and API compatibility, with a pinned install and an MCP wire-up.
Activepieces, Windmill, Node-RED, Automatisch and Huginn measured against n8n on licence, RAM, database, AI steps and the backup that breaks restores.
Bitwarden's own server wants a dozen containers and 2 GB of RAM. Vaultwarden serves the same API in one. Here is which one belongs on your VPS.
Run Langfuse on your own VPS: the real resource floor, pinned image tags, TLS, ClickHouse retention before it fills the disk, and backups that work.
Run an AI pull request reviewer on a VPS you control: self-hosted runner, diff-scoped prompts, path and size filters, inline comments, and the cost per PR.
Run the mem0 memory server on your own VPS: the real RAM floor, a Compose file that binds to localhost, TLS in front of the API, and a fully local Ollama path.
Discourse, Flarum, NodeBB and phpBB on a VPS: the RAM each really needs, the database it drags along, the spam story, and the migration path out.
Nextcloud, Seafile, Syncthing and object storage compared: sync topology, what each really costs on a small VPS, and why sync is not backup.
Your SearXNG instance can be the search backend for an AI agent. JSON API setup, the trust boundaries, and the prompt injection surface it opens.
Run the Open Connector auth gateway on your own VPS so your agents never hold a SaaS token: pinned image, TLS origin, OAuth callbacks, backups.
Restic reaches S3 and object storage on its own. Borg wants its binary on the far end and pays that back with speed over SSH. How to pick, with commands.
A staged path for learning AI agents: concepts, a loop you write yourself, tools, memory, and safety, with one thing to build at every stage.
The real cost of a Proxmox box at home versus a rented VPS: hardware and power math, what each is better at, and the hybrid setup most people run.
Two self-hosted photo servers compared: the real RAM floors, the phone apps, the maps, and the exact backup commands each one needs on a VPS.
The Compose commands you reach for every day, grouped by job: lifecycle, applying changes, logs, shells, networks, volumes, and cleanup that is safe.
Run Authentik on Docker Compose for one login across every app you host: the env values that matter, the akadmin bootstrap, and forward auth via Traefik.
What open video models really produce in July 2026, the VRAM class each one needs, what a five second clip costs on a rented GPU, and when to use an API.
Run SearXNG on your own VPS with Docker Compose: settings.yml, the limiter, nginx with TLS, and a JSON search API your own scripts can call.
Run WireGuard behind the wg-easy web UI with Docker Compose: the ports, NET_ADMIN, the sysctls that matter, and QR code onboarding for phones.
How compose.override.yaml loads on its own, what file order really merges, the ports trap that keeps a port open, and include for a dev and prod split.
Run Prowlarr, Sonarr, Radarr and qBittorrent from one Docker Compose file on a VPS, with the shared PUID, PGID and volume layout that keeps hardlinks working.
Install Listmonk on Ubuntu 24.04 with Postgres, config.toml, a systemd unit and TLS, then connect SMTP and understand what deliverability really costs.
Run your own Tailscale control server on a VPS. Install headscale from the official .deb, set server_url before you start it, then join your first node.
Tailscale is WireGuard plus a control plane. What that coordination server buys you, what it costs you, and which of the three belongs on your VPS.
Twenty things people really run on a VPS, from websites and VPNs to photo storage, game servers and AI agents, with honest RAM needs for each.
A VPS is a computer you rent. A VPN is an encrypted tunnel for your traffic. What each one does, what each costs, and how one can host the other.
Run the official Supabase Docker stack on a server you own: the secrets you must replace, what the fourteen services do, RAM needs, backups, and updates.
Run your own URL shortener on a VPS with Shlink and Docker Compose: short domain DNS, Postgres, API keys, the web client, QR codes and click stats.
AppFlowy, Outline, Docmost, BookStack and Obsidian compared on the axes that actually decide it: databases, collaboration, mobile and hosting weight.
The honest hardware ladder for running Stable Diffusion yourself: what a CPU VPS can do, what SDXL needs, ComfyUI install commands, and disk you should budget.
Run paperless-ngx on a VPS with Docker Compose: the official Postgres stack, PAPERLESS_URL, the consume folder, OCR languages, HTTPS, and backups.
Ollama is a convenience layer for one user, on CPU if needed. vLLM is a throughput engine for GPUs. Pick by workload, with the real commands for both.
Zapier and Make bill per task. Self-hosted n8n bills a flat server cost. Work a real month of volume, find the crossover, and read the n8n license.
Build a working AI agent in n8n: the AI Agent node, a Claude model credential, an HTTP Request tool, memory, a trigger, and the settings that cap cost.
Run Mealie on your own VPS with Docker Compose. Paste a recipe link and get clean steps, plus meal plans, shopping lists, nginx, TLS and backups.
Install FreshRSS on your own Ubuntu VPS with Apache, PHP and MariaDB: the release tarball, virtual host, cron refresh, and the mobile app API.
SQLite holds transactional application state. DuckDB answers analytics over Parquet and CSV. Why one VPS usually runs both, with a worked example of each.
Three self-hosted wiki apps compared on setup weight, editing model, login options and search, with a clear pick for solo admins and for teams.
Run Actual Budget on a VPS with Docker Compose: the data volume, why the browser needs HTTPS, your first budget file, bank imports, and backups.
OpenClaw, Hermes, Dify, OpenHands, and Agent Zero compared: what each agent does best, what it needs, and how to run any of them safely on your own VPS.
Install Docker Engine and Compose v2 on Ubuntu 24.04, write a real two-service compose.yml, dodge the ufw port-publishing trap, and back up volumes.
Run a Paper Minecraft server on an Ubuntu 24.04 VPS: RAM sizing, Java 21, a systemd unit, both firewalls on 25565, safe world backups, and the exact errors.
Hermes needs no GPU: the model runs elsewhere, so a small VPS is plenty. Run it as its own user with ProtectSystem=strict in systemd, and cover IPv6 in UFW.
Test outbound port 25 before installing mailcow: a block means mail never leaves the queue. Then set the DNS that stops Gmail's 550 5.7.25 PTR rejection.
A 7B model wants 8 GB of RAM and gives 4 to 10 tokens per second on CPU. Host one on a VPS and call it at 127.0.0.1:11434/v1, with port 11434 kept closed.
Self-host Rocket.Chat on a VPS with Docker Compose: the single-node MongoDB replica set it requires, TLS, backups, and every failure mode fixed.
Run Model Context Protocol servers on your own VPS so AI agents get real tools: stdio and remote HTTP transports, systemd, TLS, auth, and every failure mode.
A curated list of 25 self-hosted apps worth running in 2026, grouped by job, with what each one replaces, honest RAM and disk needs, and the gotcha to know.
Run Jellyfin in Docker on a VPS to stream your own library: block storage, media permissions, direct play vs CPU transcoding, and safe remote access.
Self-host a Bitwarden-compatible password manager with Vaultwarden and Docker on a VPS: HTTPS-first, admin token, Fail2ban, tested backups.
The honest RAM number, port 2283 behind HTTPS, the exit 137 memory kill, and why Immich v3 will not start on a pgvecto.rs database. Restore steps included.
Six containers, so budget 4 GB of RAM. Replace every secret in .env before you boot it, then create the admin account at /install before a stranger does.
Install Zabbix 7.0 LTS on Ubuntu 24.04: add the repo, import the MySQL schema, wire up the nginx frontend and agent2, then fix every error the setup throws.
A strictly hypothetical guide to the least efficient datacenter possible: one beloved server, RAID 0, heat as a strategy, and a monitor that watches itself.
Install Webmin on Ubuntu 24.04, then lock it down: SSH tunnel or IP allowlist, a real Let's Encrypt cert, two-factor, and Fail2ban on port 10000.
Run Uptime Kuma in Docker to monitor websites, ports, DNS and cron jobs, alert via email or Telegram, and publish a status page from a separate VPS.
Run n8n on a VPS with Docker Compose, Postgres, and HTTPS behind a reverse proxy: the WEBHOOK_URL and encryption-key traps, plus every error string.
Cloudron vs CasaOS vs Coolify on a VPS: install commands, TLS, backups, cost, and resource overhead compared, so you pick the right self-hosting panel.
OpenClaw is a personal AI agent that runs commands on your own server. Here is how one is built, piece by piece, and why hardening comes first.
An AI agent is a loop around a language model that can use tools. Learn the concepts behind building one on your own VPS: the loop, tools, MCP, memory.
Set up WireGuard on your own Linux VPS: key generation, wg0.conf, IP forwarding, NAT, AllowedIPs semantics, DNS, and the handshake failures that bite.
Run Nextcloud on your own VPS with Docker Compose, Postgres, Redis and a TLS reverse proxy, plus the backup and upgrade steps that keep your data recoverable.