Back to Documentation

Building Your Wiki

This guide shows, in practice, how your project's Wiki is built and how to navigate it. Unlike other sections, the Wiki is not filled in through a "new item" form: it is built by an AI assistant connected to the project and read and navigated through the interface. By the end, you will know how to enable the Wiki, understand how nodes and links are formed, enrich the network with concepts and relations, and explore the knowledge graph.

Before you begin: The Wiki requires a paid plan (starting with Pro) and must be enabled in the project. You will also need an assistant connected via MCP — see Connecting Sutram to Claude. If you haven't read it yet, start with the concept in Wiki and Knowledge Graph.

Step 1: Enable the Wiki in the project

The Wiki only appears after it is activated by the owner:

  1. As the project owner or admin, open the project settings
  2. Enable the Wiki option
  3. The Wiki card starts appearing on the project's home page

Step 2: Understand how the Wiki is built

The Wiki does not fill itself, and the web interface has no "new node" form: all node creation and editing goes through the Sutram MCP Server, that is, through an AI assistant connected to the project. The interface is where you navigate, search, and read what has been built.

What forms from there:

  • Record mirror nodes — when you ask to mirror a Record, the assistant creates the corresponding node and copies the document's full text into it, bringing that content into the search and the graph.
  • Concept nodes — concept or synthesis pages that do not correspond to a file, created on demand. When creating one, Sutram compares the name against the project's catalog and reuses a similar node instead of duplicating.
  • Mentions create the network — when a node's text uses [[slug]] to cite another, that link becomes a graph edge. Citations to something that does not yet exist become stubs (ghost nodes) and are recorded as visible pending items; as soon as the target is created with that slug, the mention resolves on its own.

In other words: you guide the construction by talking with the assistant, and the structure emerges from the links.

Step 3: Enrich with concepts, syntheses, and relations (via AI)

In practice, creating concept nodes, writing syntheses, and defining typed relations are done by asking the AI assistant connected to the project through the Sutram MCP Server (for example, Claude). Some examples of what you can ask:

  • "Create a wiki node for the concept 'Customer onboarding' with a synthesis of our process."
  • "Link this document to the 'Refund Policy' node with the 'depends on' relation."
  • "Bring this Record's full text into the corresponding mirror node, so it enters the search."

Behind the scenes, the assistant uses the MCP's Wiki tools — creating and updating nodes, defining mentions and typed relations, and syncing the source text. The full reference for these tools is in the Sutram MCP Server Guide.

Tip: start by asking the AI to mirror the Records that matter most and to write short syntheses. Then evolve toward typed relations where the "how" of the connections matters.

Step 4: Navigate the Wiki

Open the Wiki card on the project's home page. It has three sections:

  • Concepts — the list of concept nodes (and orientation nodes), each with its backlink count
  • Documents — the mirror nodes of the project's documents
  • Graph — the visualization of the relations between the nodes

Click any node to open its page. On it you'll find:

  • The node's synthesis, with the [[slug]] mentions rendered as clickable links (the pending ones appear marked as broken)
  • The full text, when the node mirrors a document
  • The ego-graph — a radial map of the node's immediate neighborhood
  • The relations panel and the backlinks ("Mentioned in"), to jump to everything that connects to it

Step 5: Explore the graph

In the Graph section, you see the project's knowledge as a network:

  1. Search or filter by node name or type to focus on what interests you
  2. Click a node to open it or view its relations
  3. Navigate the clusters to drill down into related themes
  4. Rearrange the layout by dragging the nodes, and save or discard the arrangement

Use the graph to answer questions a list cannot: which concepts are central? What is isolated and might need context? How do two areas of the project connect?

Curation Tips

  • Stable names — the slug is the node's address; standardize names to avoid duplicates and broken mentions.
  • Lean syntheses — the entry orients; the detail is already in the document's full text.
  • Early mentions — cite a concept with [[slug]] even before documenting it; the pending item becomes a visible reminder.
  • Relations with a verb — "revokes", "amends", "depends on" tell a story that a loose link does not.
  • Review the pending items — broken mentions are a good map of what still needs to be documented.

Frequently Asked Questions

Q: I don't use AI. Do I still have a Wiki?

A: No. Today the only way to add or modify Wiki nodes is through the Sutram MCP Server — the web interface does not offer that functionality. Without a connected assistant, the Wiki stays empty. Once built, however, any participant can navigate, search, and read everything through the interface, with no AI needed.

Q: How do I create a concept node without AI?

A: It is not possible through the interface. Connect an AI assistant to the project (see Connecting Sutram to Claude) and ask it to create the node — this is the path by which the Wiki is built and curated.

Q: Why does a node appear "broken" or empty?

A: It is a stub — a node created by a mention to something not yet documented. As soon as the real target (a Record or a concept node) is created with that slug, the mention resolves and the node gains content.

Q: Who can build and who can only read the Wiki?

A: Navigating, searching, and reading are open to any participant, including viewers. Creating and editing nodes, syntheses, and relations requires an editing role (Owner, Admin, or Member).

Next Steps

  • Records and Metadata: structured data in Sutram — how to structure the data that feeds the Wiki's mirror nodes
  • Sutram MCP Server Guide — the reference for the Wiki tools used by the AI assistant
  • Wiki and Knowledge Graph — the concept behind what you just built

Document Version: 1.0 Last Updated: July 2026 Author: Sutram Development Team

Back to Documentation