§ 01
Before you pick an app
You want to know what an app needs before you offer it to a customer.
| Your server is checked first | Each deploy checks your server against the app's minimum size, shown in the lists below. |
|---|---|
| Some apps take their AI route at deploy | Six AI apps do: Hermes Agent, Hindsight, LiteLLM Proxy, Open WebUI, OpenClaw and OpenHands. The routes each one offers depend on the app. Other apps take an AI key in the deploy form or set their AI provider inside the app. |
| New apps start private | Only people signed in to your panel can open them, and some templates declare their own exceptions. Features such as incoming webhooks need the app to be public. Publishing can't be undone yet. |
| Some apps make their first visitor the administrator | Set up the admin account while the app is still private. |
Note
The catalog lists what we package. Check each app with your own setup before you offer it to customers.
§ 02
AI apps
| App | What it does | Minimum |
|---|---|---|
| AnythingLLM | Chat with your own documents and run AI agents. Supports several AI providers and user permissions. | 1 core, 2 GB |
| Bolt.diy | Build and edit web apps in the browser from plain-language prompts. You choose the AI provider. | 2 cores, 2 GB |
| ContextForge MCP Gateway | IBM's MCP gateway. One endpoint with sign-in brings many MCP servers, tools and APIs together. | 1 core, 1 GB |
| DeerFlow | ByteDance's research agent. It plans long tasks, writes code and produces reports, slides and web pages. Bring your own AI key. | 2 cores, 4 GB |
| Dify | A visual builder for chatbots, document search and AI agents. AI providers are set up inside Dify. | 2 cores, 6 GB |
| Hermes Agent | Nous Research's agent, with a web dashboard for multi-step tasks. AI route chosen at deploy. Chat channels connect in its own dashboard. | 1 core, 2 GB |
| Hindsight | Long-term memory for AI agents, through a REST API or MCP. It needs an AI route at deploy. | 2 cores, 2 GB |
| IronClaw | NEAR AI's agent runtime, inspired by OpenClaw. Bring your own AI key. One user per app. Its sandbox for shell tools is switched off in this version. | 2 cores, 2 GB |
| Langflow | A drag-and-drop builder for AI agents and document search. AI providers are set up inside the app. | 1 core, 2 GB |
| LibreChat | One chat app for many AI models, with roles, search and file uploads. AI providers are set up inside the app. | 1 core, 2 GB |
| LiteLLM Proxy | One OpenAI-compatible endpoint for many AI providers, with keys, budgets and spend tracking. AI route chosen at deploy. | 1 core, 1 GB |
| LocalAI | A local AI server with an OpenAI-compatible API, for text, images, audio and embeddings. Runs without a GPU. | 2 cores, 4 GB |
| Ollama | Runs open models such as Llama, Mistral and Gemma on your server. A GPU isn't required, and speed depends on your hardware. | 2 cores, 4 GB |
| Open WebUI | A chat interface for local and OpenAI-compatible models, with several users, document upload and web search. AI route at deploy: a local model. | 1 core, 1 GB |
| OpenClaw | Build and run AI agents that use tools, through a gateway with its own password. AI route chosen at deploy. Review its tool settings inside the app. | 2 cores, 3 GB |
| OpenHands | A coding agent that writes code, runs commands and tests its work. AI route chosen at deploy. It has no access to the server's Docker. | 2 cores, 4 GB |
| Paperclip | An open-source app for running a team of AI agents, each with its own role. Its admin account is created with a command after deploy. | 1 core, 1 GB |
| RAG API with ChromaDB | ChromaDB, a vector database for document search. Your own app adds the upload step and the AI model. | 1 core, 2 GB |
| SearXNG | A metasearch engine that combines results from many others, and sends your searches to them. AI apps often use it for web search. | 1 core, 512 MB |
| text-generation-webui | A web interface to run and test open models locally. Its default build is made for NVIDIA GPUs. | 2 cores, 4 GB |
| Vane, formerly Perplexica | An AI answer engine that searches the web and cites its sources. SearXNG is included. | 2 cores, 2 GB |
| vLLM | An OpenAI-compatible engine for serving models under heavy use. It needs a GPU. | 4 cores, 8 GB |
| Whisper ASR | Speech to text and translation through a REST API. Based on OpenAI's Whisper. | 2 cores, 4 GB |
Note
OpenClaw can also link one WhatsApp number by a QR code on its page in the panel. It allows one sender and keeps groups switched off. It is separate from the WhatsApp Business API.
§ 03
Databases
| App | What it does | Minimum |
|---|---|---|
| MariaDB | An open-source relational database that can replace MySQL. | 1 core, 512 MB |
| MongoDB | A document database for flexible, JSON-like records. | 1 core, 512 MB |
| MySQL | An open-source relational database. | 1 core, 512 MB |
| PostgreSQL | An open-source relational database, version 16. | 1 core, 256 MB |
| Qdrant | A vector database for semantic and document search, with a web dashboard. Its API key is generated at deploy. | 1 core, 1 GB |
| Redis | An in-memory store for caching, sessions and messages. | 1 core, 128 MB |
Caution
The panel backs up PostgreSQL databases and refuses other engines. It can't restore a database backup on its own. Plan backups for other engines another way.
§ 04
Developer tools
| App | What it does | Minimum |
|---|---|---|
| Workspace with Claude Code | VS Code in the browser, with Anthropic's Claude Code in the terminal. Codex, OpenCode and Pi are installed too. | 2 cores, 4 GB |
| Codex | VS Code in the browser, with OpenAI's Codex ready in the terminal. | 2 cores, 4 GB |
| OpenCode | VS Code in the browser, with the OpenCode coding agent. | 2 cores, 4 GB |
| Pi | VS Code in the browser, with Pi, a small coding agent. Pi runs every command it chooses without asking, inside its container. | 2 cores, 4 GB |
| code-server | VS Code in the browser, on your server. | 1 core, 1 GB |
| Gitea | A light Git service with code review, issues and CI. Git clients and webhooks reach it once it is public. | 1 core, 256 MB |
| n8n | Visual workflow automation with many ready-made integrations. Incoming webhooks work once the app is public. | 1 core, 512 MB |
| Docker Registry | A private registry for container images. It has no login of its own, so a public registry lets anyone push and pull. | 1 core, 256 MB |
| RustFS | S3-compatible object storage in one container. Its access keys are generated at deploy. | 1 core, 256 MB |
| Uptime Kuma | Monitoring with status pages and many notification services. Public status pages need the app to be public. | 1 core, 128 MB |
Note
The four coding workspaces use the developer's own subscription or key. The panel's AI routes are not available to them. Files and sign-ins are kept, and running sessions end when the workspace restarts.
§ 05
Web apps
| App | What it does | Minimum |
|---|---|---|
| BookStack | A wiki for documentation, organised in shelves, books, chapters and pages. Change its default admin login before you publish it. | 1 core, 1 GB |
| Directus | A headless CMS with REST and GraphQL APIs. Its admin app sits on top of a SQL database. | 1 core, 1 GB |
| Ghost | A publishing platform with memberships and newsletters. | 1 core, 1 GB |
| Karakeep, formerly Hoarder | A bookmark manager that tags and organises saved links with AI. The AI provider is set inside the app. | 1 core, 1.5 GB |
| Mattermost | Team chat that runs on your server. | 1 core, 1 GB |
| Nextcloud | File sync, calendars, contacts and shared work in one suite. | 1 core, 1 GB |
| Plausible Analytics | Website analytics without cookies. | 2 cores, 2 GB |
| Stirling PDF | PDF tools in the browser for merging, splitting, converting and signing. It also reads text with OCR. | 1 core, 1 GB |
| Supabase | A PostgreSQL database with sign-up and login, instant APIs, realtime and file storage. All its keys are generated at deploy. | 2 cores, 2 GB |
| Umami | Simple website analytics without cookies. Change its default login before you publish it. | 1 core, 512 MB |
| WordPress | The widely used system for websites and blogs. Run its install wizard, then publish it. | 1 core, 512 MB |
§ 06
Missing an app?
You want to offer an app that isn't on this list.
Deploy it from a public Git repository, or bring a Docker Compose file. Some repositories need changes before they run. You can also tell us which app you need.
App names and trademarks belong to their owners. Each app is made by its own developers and comes with its own licence and terms. Listing an app here doesn't mean its makers endorse AI Admin Panel.