Wikantik is an agent-grade knowledge base — a wiki that humans read and AI agents can natively search, retrieve, and cite. One Markdown source of truth, two kinds of reader, the same content and permissions for both.

As a human, browse in-depth, interlinked research articles organized into topic clusters — each cluster has a hub page that introduces the topic and links to its sub-articles. Search in plain language (hybrid keyword + semantic retrieval), and explore how ideas connect through the page graph of wikilinks and the knowledge graph of entities extracted from the content.

As an AI agent, connect over the Model Context Protocol: the read-only /knowledge-mcp server retrieves and assembles ranked, citation-bearing context bundles and traverses the knowledge graph, while /wikantik-admin-mcp authors and curates pages — and the whole corpus is queryable as linked data over SPARQL. Agents go through the same save pipeline, permissions, and retrieval index as the human UI; see Wikantik System Architecture for how it all fits together.

New here? Start with About for the manifesto, or the Wikantik Platform Hub for the platform deep-dives. Power users: the Knowledge Graph Admin guide. You can also read the site blog.

Article Clusters

Finance & Retirement

Geopolitics & Conflict

Technology

Agent Cookbook

Hobby Woodworking

Warehouse Automation

Berlin History

Immigration

Wikantik Platform

About This Wiki

Powered by Wikantik. Articles are written as AI-assisted research — drafted with Claude, reviewed and edited by a human, and published via the MCP API. Content is organized using YAML frontmatter metadata for programmatic discovery and cross-referencing.