Nuclino MCP. Query and write directly to your corporate wiki.

Look, you're connecting your AI agent straight into the guts of your company knowledge base with Nuclino. This isn't just another read-only API wrapper; it gives your agent full read and write access to everything—every team, every workspace, every single document. Your agent can act like an employee who actually knows where things are stored.

You don't have to copy/paste or manually search anymore.

First off, you gotta map out the landscape. To figure out what docs exist and how they’re organized, your agent uses list_teams to see every team attached to your account; then it maps those teams using list_workspaces to show all the isolated project spaces within them. You can drill down further by running list_collections to check document groupings, or you can map out the entire content structure by calling list_items to get a list of every standard wiki page title and its UUID.

To see what data points are available across the board—the custom fields they use for things like project IDs or version numbers—your agent runs list_fields. If you need to know who’s on which team, it uses list_users after identifying a team via list_teams.

When you're hunting down info, your agent has two ways. You can run a broad semantic search using search_items, and that lets the AI client find relevant documents across the entire organization's knowledge base based on keywords or context. If you know exactly what document it needs, running get_item retrieves the full raw Markdown content and all the metadata for that specific item.

You can also check what physical attachments—the actual binary files or PDFs—are linked to any given piece of documentation by calling list_files.

If your agent needs to write something, it's got its tools ready. To build a brand new wiki page from scratch based on natural language prompts, it executes create_item, generating a permanent document item within the target workspace. If an existing draft is wrong or incomplete, running update_item lets the agent overwrite or append fresh content to that established item right away.

And yeah, if documentation gets totally outdated and nobody needs it anymore, your agent can remove that structural waste using delete_item, though remember it'll probably ask you for confirmation first because it’s a serious move.

It's all about control and knowing where stuff lives. To understand the full scope of data available, your agent runs through these steps: listing teams with list_teams, mapping out workspaces with list_workspaces, then checking collections with list_collections. If it needs to check the specific properties attached to any document, it uses list_fields and can see what files are sticking to a single item using list_files.

Basically, your AI client gets full visibility: it searches for documents via search_items; it manages the content lifecycle by creating new items with create_item, updating old ones with update_item, and deleting junk with delete_item; it maps the structure by listing teams (list_teams), workspaces (list_workspaces), collections (list_collections), and fields (list_fields); and it reads anything—from full markdown content via get_item to just a list of page titles using list_items.

You've got everything it needs right there.