Overview

Hjarni is a hosted knowledge-base service with a built-in MCP server, launched May 2026. It offers zero-setup operation: Claude and ChatGPT can share the same brain through MCP. In the comprehensive report it is the “Hosted (Zero-Setup)” pattern (Pattern D), aimed at non-technical users.

Key Facts

  • Type: hosted MCP wiki service (zero local setup)
  • MCP: built-in MCP server so multiple LLMs (Claude + ChatGPT) share one brain
  • Audience: non-technical users, teams wanting zero maintenance
  • Launch: May 2026 (second wave of LLM Wiki infrastructure)
  • URL: https://hjarni.com

Features

Core Architecture

  • Hosted service — zero local setup, runs entirely in the cloud
  • MCP-native — built-in Model Context Protocol server
  • Multi-LLM sharing — Claude + ChatGPT share the same brain via MCP
  • Zero maintenance — no terminal, no config, no updates to manage

Ingest & Processing

  • Web-based ingest — add sources through the hosted interface
  • URL capture — save web content directly

Search & Retrieval

  • MCP search — hybrid search via Model Context Protocol
  • Cross-LLM consistency — same knowledge base accessible from multiple LLMs

Integration

  • Claude + ChatGPT — multi-LLM support via MCP
  • Zero-setup — no installation, no configuration
  • Teams — shared knowledge base for teams

Data & Storage

  • Cloud-hosted — managed infrastructure
  • MCP protocol — standardized agent access

Relationships

  • “powered-by” model-context-protocol — its core mechanism is a hosted MCP server
  • “compatible-with” llm-wiki — hosted implementation of the pattern
  • “sibling-of” lucasastorian-llmwiki — also offers a hosted option, but Hjarni is MCP-native
  • “contradicts” cognee — contrast: Cognee is self-hosted graph; Hjarni is hosted MCP

Implications

Hjarni is the “remove the manual entirely” option the user has been pushing for — no terminal, no config. The trade-off (already noted for hosted tools) is vendor lock-in and less local control, the same caveat raised earlier about nashsu-llm-wiki.

Sources

^[raw/articles/llm-wiki-okf-comprehensive-report-2026.md]