Architecture

Understanding RA-H's AI system

Overview

RA-H uses a single built-in assistant plus an MCP server for external agents. The system is built around three core concepts:

  • Nodes - Your knowledge items (papers, ideas, people, projects)
  • Edges - How things connect with explanations
  • Dimensions - Flexible tags that emerge from your content

Core Elements

The three core elements: nodes, edges, and dimensions

Nodes

Knowledge items stored in the database. Each node can be a paper, idea, person, project, video, tweet, or any other piece of information. Nodes have:

  • Title and notes (your thoughts)
  • Description (AI-generated grounding)
  • Dimensions (multi-tag categorization)
  • Metadata (structured information)
  • Embeddings (for semantic search)
  • Links (to external sources)
  • Chunk (full source content)

Edges

Directed relationships between nodes. Edges capture how your knowledge connects with explanations - not just simple links, but contextual relationships like "this paper inspired this idea."

Dimensions

Multi-select categorization tags. Unlike rigid folders, nodes can have multiple dimensions.

Built-In Assistant

  • Runtime: RA-H
  • Model: GPT-5 Mini
  • Behavior: Multi-step, direct graph access, grounded by context + skills
  • Writes: No approval gate in the current product
  • Best for: Day-to-day graph work, synthesis, cleanup, and capture

External Agents via MCP

Claude Code and other MCP clients connect to the same graph through the standalone MCP server. They can read and write graph data and use the shared skills system.

Typical flow:

  1. Connect the MCP server.
  2. Call getContext first for orientation.
  3. Search before creating.
  4. Read skills when procedural detail matters.
  5. Say let's get started on first run to trigger onboarding.

Graph Quality

Graph quality is a retrieval feature, not just data hygiene:

  • explicit node descriptions make retrieval and grounding better
  • good dimensions improve search and browse flows
  • explicit edge explanations create usable graph structure for AI reasoning

Context Management

RA-H keeps context grounded with:

  • Hub nodes — Top connected nodes included automatically
  • Focused node — Current working node with description/notes preview
  • Skills — User-authored markdown instructions loaded on demand