Practical overview
When it fits
Use Arkon when a team wants to curate organizational documents into a reviewed, traceable wiki and expose the resulting context to AI clients with department- and role-based access boundaries.
Who it is for
- Teams organizing policies, procedures and internal documentation for AI-assisted retrieval.
- Organizations that need separate knowledge scopes and role-based access across departments.
- Administrators able to deploy a Docker-based service and configure external AI providers.
Documented uses
- Compile internal documents into linked wiki pages with traceable source references.
- Search and browse team knowledge using full-text and semantic search.
- Provide Claude and other AI clients with scoped access to wiki pages and source material through MCP.
- Review proposed wiki changes and manage drafts, approvals and page history.
Requirements & limits
The repository describes Arkon as open source and documents self-hosted Docker deployment. Setup requires an API key from a supported AI provider.
- Arkon is a multi-container service; the README lists seven Docker containers and recommends 2 vCPU, 4 GB RAM and 40 GB SSD for a starter setup.
- The wiki compilation and search stack uses PostgreSQL with pgvector, Redis and MinIO alongside FastAPI and a Next.js frontend.
- AI inference is handled through external Anthropic, Google or OpenAI APIs, so provider selection and credentials are part of setup.
- Access boundaries can be organized by department and global scope, with roles including Viewer, Contributor, Editor and Admin.
- The documented Claude connection uses an MCP custom connector and OAuth 2.1 with PKCE.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up Arkon with Docker
Launch the self-hosted Arkon portal and configure its AI providers.
Before you start
- Docker and Docker Compose
- An API key from Anthropic, Google or OpenAI
- A server environment; the README recommends 2 vCPU, 4 GB RAM and 40 GB SSD for a starter deployment.
Steps
Clone the Arkon repository and change into its directory.
Copy
.env.docker.exampleto.env.dockerand set the secret key, administrator credentials, and PostgreSQL and MinIO secrets.Run the command below.
docker compose --env-file .env.docker up -d --buildOpen the portal at
http://localhost:3119, sign in as administrator, and configure embedding, LLM and vision models with provider API keys.
Expected outcome
The Docker deployment starts the Arkon portal, where an administrator can select AI models and provide their API keys.
Guide limitations
- The documented setup requires external AI-provider credentials.
- The README points to the separate Setup Guide for development mode and the full environment-variable reference.
Connect Claude to an Arkon MCP server
Add an existing Arkon server as a Claude connector and sign in.
Before you start
- A running Arkon server with its MCP endpoint reachable at the server URL plus
/mcp - Arkon login credentials
- Claude Desktop or Claude.ai
Steps
In Claude Desktop or Claude.ai settings, add a custom connector named
Arkon.Set its URL to
https://your-arkon-server/mcp, replacing the example host with the Arkon server address.Select Connect and sign in through the browser window opened by Arkon.
Expected outcome
Claude is connected to Arkon’s MCP tools for searching and reading knowledge wiki content, subject to the connected user’s permissions.
Guide limitations
- The instructions describe Claude Desktop and Claude.ai; they do not provide setup steps for other clients.
- Access to knowledge is constrained by the user's assigned scopes and permissions.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 16 October 2026
- Original website title
- nduckmink/arkon · Official homepage title · retained in its source language