# Connectors Source: https://docs.getprimo.com/ai/connectors Connect external tools to Primo AI by registering Model Context Protocol (MCP) servers. Connectors give [Primo AI](/ai/primo-ai) access to tools outside the platform. A connector is a Model Context Protocol (MCP) server that you register by URL. Once connected, the assistant can call that server's tools during a conversation — for example, to look up a page in Notion or act in another system. ## When to use a connector Add a connector when you want the assistant to reach beyond Primo's own data: * **Query another system** — pull information from a tool your team already uses. * **Act in another system** — let the assistant perform tasks in a connected service. * **Extend the assistant** — bring in any tool that speaks the Model Context Protocol. If you only need the assistant to work with Primo data or your company's own instructions, you don't need a connector — use [knowledge base items](/ai/knowledge-base-items) instead. ## Prerequisites * The **AI agent write** permission to add, re-authenticate, or delete connectors. * The URL of an MCP server that supports OAuth authentication. ## Add a connector Open **Settings**, select the **AI** section, then the **Connectors** tab. The connector dialog opens. * **Name** — a label for the connector, for example `Notion`. * **Server URL** — the MCP endpoint, for example `https://example.com/mcp`. Expand **Advanced settings** only if your MCP server does not support Dynamic Client Registration. Register Primo in the provider's OAuth console, then enter the **Client ID** and **Client secret**. Leave the secret empty for public clients that use PKCE. Primo redirects you to authenticate with the MCP server. After you authorize access, the connector returns to Primo and its status becomes **Connected**. ## Connector statuses | Status | Meaning | | -------------------------- | ------------------------------------------------ | | **Pending authentication** | Created, but authentication is not yet complete. | | **Connected** | Authenticated and available to the assistant. | | **Error** | The connection failed. Re-authenticate to retry. | ## Manage connectors From the Connectors tab you can: * **Re-authenticate** — refresh access when a connector expires or errors. * **Delete** — remove a connector and its stored credentials. ## How Primo AI uses connectors When a conversation needs an external tool, the assistant selects the relevant one from your connected connectors and asks you to approve the action before it runs. Open Primo AI and use the **Connectors** control in the message box to choose which connectors are active for a conversation. See [Primo AI](/ai/primo-ai). # Knowledge base items Source: https://docs.getprimo.com/ai/knowledge-base-items Teach Primo AI your company's knowledge, policies, and reusable tasks by creating and publishing knowledge base items. Knowledge base items are the instructions and reference material [Primo AI](/ai/primo-ai) draws on when it answers a request. Think of each item as something you would train a new team member on: a policy to follow, a procedure to run, or a fact to remember. In the dashboard, knowledge base items appear as **skills** under the Knowledge base tab. Each item can capture reference knowledge (what is true) or behavioral guidance (how to act) — or both. ## When to use a knowledge base item Create an item whenever the assistant should behave consistently with the way your company works, not just its general knowledge. For example: * **Encode a policy** — "Devices must run the latest major OS within 30 days of release." * **Standardize a procedure** — the exact steps your team follows to offboard a departing employee. * **Record company facts** — approval thresholds, naming conventions, or who owns which tool. * **Reuse a task** — a routine the assistant should be able to run the same way every time. If it tells the assistant **how to act**, write it as a procedure. If it tells the assistant **what is true**, write it as reference. A single item can do both. ## Open the knowledge base Open **Settings**, then select the **AI** section. The tab groups items by status: **Published**, **Pending changes**, and **Unpublished**. ## Create an item A new item opens in the editor. Give it a short, specific name (up to 120 characters). The assistant uses the name to decide when the item is relevant. State **when** the assistant should use this item, in one line (up to 1024 characters). The assistant reads the description to decide whether an item applies to a request, so describe the trigger, not the content — for example, "Use when an employee reports a lost or stolen device." Write the knowledge itself in Markdown. Describe the policy, procedure, or facts in plain language, the same way you would brief a colleague. Changes save automatically. The editor shows **All changes saved** once your work is stored. ## Publish an item An item only reaches the assistant after you publish it. Drafts stay private. * Click **Publish** to make a new item available to Primo AI. * Edit a published item and its status becomes **Pending changes**. Click **Publish changes** to push the update. * Click **Unpublish** to remove an item from the assistant while keeping the draft. ## Item statuses | Status | Meaning | | ------------------- | --------------------------------------------------------- | | **Unpublished** | Draft only. Not visible to the assistant. | | **Pending changes** | Published, but the draft has edits that are not live yet. | | **Published** | Live and available to Primo AI. | ## How Primo AI uses your items When you chat with the assistant, it scans your published items and pulls in the ones relevant to your request, then follows their guidance in its answer. The assistant only uses published items — it ignores drafts and unpublished items. Primo shares items across your company, so every teammate's assistant benefits from what you publish. ## Best practices Keep each item focused on a single policy or procedure. Small, specific items are easier for the assistant to match and apply. The assistant relies on the name and description to decide when an item applies. Make them specific. Write as if briefing a new colleague. Avoid internal jargon the assistant can't interpret. Publish early, watch how the assistant uses the item, and edit as needed. Republish to push changes. ## Feature-managed items A Primo feature creates and maintains some items, such as an equipment or purchasing policy. These show a **Used by** label, and you can't edit or delete them from the knowledge base — manage them from the feature that owns them. # Primo AI Source: https://docs.getprimo.com/ai/primo-ai Chat with Primo AI, a context-aware assistant that helps you search, understand, and act across the platform.