Does an XWiki MCP server need an admin account?

No. A regular user account with view access to the relevant spaces is enough. A dedicated account with narrowly scoped access is safer than using an admin account, which would give the assistant access to all spaces and administrative functions.

Can an AI assistant edit XWiki pages over MCP?

Only if the MCP server exposes write tools and the connected account has edit permission. For most workflows, read-only access is the right starting point. Wiki content typically reflects reviewed decisions; letting an assistant write directly bypasses that review.

How do I handle confidential wiki spaces?

Create the MCP account with view access only to the spaces needed for the task. Verify the account cannot see confidential spaces before connecting the assistant. Test by logging into XWiki with that account and confirming that confidential spaces are not visible.

What is the XWiki REST API version to target?

Check the XWiki version your organization runs and the API version it supports. XWiki's REST API documentation on xwiki.org describes the available endpoints for each version. The server documentation should state which XWiki version it was tested against.

Can I connect both XWiki and a task tool to the same AI assistant?

Yes. Most AI assistants support multiple MCP connections at the same time. Name each server clearly in the client configuration so the assistant uses the right tool for each request. A connection to XWiki for documentation and a connection to fenbs for tasks and decisions work well together.

XWiki MCP Server: Connect AI to a Team Knowledge Base

An XWiki MCP server lets an AI assistant search and read wiki pages during a work session. Here is how to evaluate connection options, check page-access permissions, and use fenbs to keep project decisions and follow-up work linked to your team's source documentation.

October 4, 2026Updated October 21, 20264 min read

An XWiki MCP server gives an AI assistant the ability to search wiki pages, read page content, and navigate the space structure during a work session. Instead of telling the assistant what your team's process documentation says, it can read the relevant page and reason from the actual content. Tools like fenbs keep the project decisions and follow-up tasks linked to the source documentation, so a person reviewing a decision can follow it back to the wiki page that informed it.

XWiki is an open-source wiki platform used by organizations that want to host their knowledge base on their own infrastructure. It has a REST API that covers pages, spaces, attachments, and search, which provides a solid foundation for an MCP connection.

What an XWiki MCP server can expose

The XWiki REST API covers page content, page history, space listings, child pages, attachments, and full-text search. An MCP server wraps those endpoints as named tools. Common operations include searching the wiki for a term, reading the current content of a specific page, listing child pages under a parent, and reading page metadata such as last editor and modification date.

Write tools, when exposed, can create or update wiki pages. For most AI-assisted work, read-only access is the right starting point. Content that appears in a wiki has typically gone through a review; letting an assistant write directly bypasses that review step.

Evaluating an XWiki MCP server

Search the MCP server directory and community repositories for XWiki MCP tools. When evaluating one, check the XWiki API version it targets. XWiki has had multiple REST API versions; confirm the server works with the XWiki release your organization runs.

Also check the authentication method. XWiki supports basic authentication and token-based authentication. Token-based authentication is safer for MCP connections because it does not expose account passwords. Check whether your XWiki version supports API tokens in the user profile settings.

Checking page access before connecting

XWiki's access control is per-space and per-page. A user account may have view access to some spaces and not others. The MCP connection inherits the access of the account it uses. Before connecting an assistant, confirm that the account can see the spaces and pages needed for the task, but not spaces that contain confidential information the assistant should not reach.

Create a dedicated XWiki account for the MCP connection with view access limited to the relevant spaces. If your organization's XWiki instance contains HR records, financial data, or customer information in separate spaces, ensure the MCP account does not have view access to those spaces.

Record the access decisions in a shared place. A note stating which spaces the assistant may read and which require human mediation tells every team member the intended boundary without them having to check XWiki's access settings directly.

Keeping decisions linked to source documentation

The value of a wiki as a source for decisions comes from its traceability: you can read the process page that a decision was based on, then check whether the page has been updated since the decision was made. A decision record that includes a link to the relevant wiki page is far more useful than one that summarizes the page without attribution.

When an assistant reads an XWiki page and produces a decision or a recommendation, save the wiki page URL alongside the decision. If the page changes later, the link shows that the current content may differ from what was in place when the decision was recorded.

How fenbs helps

Link project decisions to the documentation behind them

fenbs keeps Links and listings and Decisions and rules in one shared place. Save each XWiki page URL alongside the decision or task it informed. When the decision is reviewed later, the link goes back to the source. Connected AI assistants read the decisions before starting work, so they know what was already agreed and where the supporting documentation lives.

Create a free account

Free for one person · No credit card · Works with Claude, ChatGPT, Cursor and more

Practical workflows with a wiki-connected assistant

Ask the assistant to read the process page for an unfamiliar workflow before helping you follow it. Or ask it to search the wiki for the current approved approach to a recurring task. The result is more accurate than asking the assistant to recall from training data, because it reads what your team actually documented.

After following a process, record any gaps or corrections in a task rather than immediately editing the wiki. That gives a person the chance to review the proposed change before it becomes the documented standard. See mcp-server-configuration for general advice on authentication and tool scoping across MCP clients.

Confirm the page title the assistant retrieved matches the page you intended before treating its summary as the current standard. XWiki spaces can contain multiple pages with similar titles, and a query that returns the wrong one gives the assistant inaccurate context. Check the page URL or last-edited date when the result matters.

How to do this with fenbs

Create a free account and set up a project for the team or workstream that uses XWiki as its documentation base.
Add Links and listings for the key wiki spaces and pages: the space URL, the page title, and what it covers.
Connect your AI assistant over MCP from Integrations. Ask it to read the project context before querying XWiki, so it knows which pages are relevant and which decisions have already been recorded.
When the assistant reads a wiki page that informs a decision, record the decision in fenbs with the XWiki page URL as a linked source.
When a wiki page changes, revisit the decisions linked to it and note whether they still hold or need updating.