Skip to main content
Role briefs, hiring criteria and company research often live in other tools. A connector lets the Sourcing agent use that information when you ask, without copying it into each search. Every connector works the same way and handles data the same way. They differ only in how you set them up and what the agent is allowed to do.
Connectors bring your tools into Metaview. To use Metaview from an AI tool such as Claude or ChatGPT, see MCP Integration Overview.

​
Available connectors

The ready-made connectors are preconfigured versions of the same connector that a custom connector uses. Everything on this page applies to all of them. Each tool’s own limits and usage charges still apply. For example, some Atlassian requests consume Rovo credits.

​
How it works

Connectors use the Model Context Protocol (MCP), a standard way for an app to offer an AI agent a list of tools, such as “search documents” or “get open roles”. Each tool has defined inputs and outputs. An MCP server is the service that hosts those tools. The providers of the ready-made connectors host their own; for a custom connector, you supply the server. When you select a connector in a search, the agent can call the tools its server offers:
  1. You ask the agent to use the connector, for example “Find the Head of Engineering brief in Box.”
  2. The agent calls a tool on the server, sending the inputs that tool accepts, such as a search term or role ID.
  3. The server responds with whatever that tool is built to return.
  4. The agent uses the response to guide the search, such as the must-have criteria from a role brief.
The agent can only call the tools the server exposes, and only sees what those tools return. Metaview doesn’t connect directly to your databases or files.

​
What connectors are used for

Connectors give the agent hiring context, such as role briefs, intake notes and target-company lists. For example: There is no minimum number of documents or records and no required list of fields. One useful role brief can be enough.

​
Use a connector in Sourcing

Open Sourcing and use the Connectors menu beside the message composer to select the connector for your search. Then tell the agent which information to use and what you want to do with it. Document titles, folder names and links help it find the right content.

​
Data access and security

​
What the agent can access

Access is set by three things you choose: which connector you use, the account or API key you connect, and the connector you select for each search.
  • Scope comes from the connected account. The agent can reach only what that account can. How granular permissions can be depends on the provider, or on how your own server is built. Some providers, such as Airtable, also let you choose what to grant when you sign in. Where possible, connect an account or API key limited to the folders, pages or records you want the agent to use.
  • Read and write access depends on the connector. See Available connectors.
  • Information is retrieved on request. Connecting a tool doesn’t import everything the account can access. The agent calls a tool only when you ask it to use that connector, and requests what it needs for your search.

​
What data moves between Metaview and your tool

Data moves in both directions:
  • From your tool to Metaview. The agent receives everything the server includes in a tool’s response.
  • From Metaview to your tool. The agent can include details from your search in the inputs it sends when calling a tool.
Connectors don’t need candidate personal information to be useful. If a response does include it, such as names, email addresses or phone numbers, Metaview retrieves it along with the rest of the response and doesn’t automatically remove it. To keep personal information out, connect an account that can’t reach it or, for a custom connector, limit what your tools return to role and company information.

​
Storage and logging

Information a connector returns is stored by Metaview as part of the search conversation. Metaview also logs the tool calls the agent makes, including the inputs it sends. Disconnecting a connector stops further requests, but it doesn’t remove information already returned in a search.

​
Your responsibility

You are responsible for deciding what is appropriate to expose through a connector. Before you connect a tool, confirm that:
  • the account or API key has only the permissions you intend the agent to have;
  • the content is suitable to use for sourcing under your company’s policies and applicable data protection requirements, including a legitimate basis for any personal information; and
  • you are comfortable with the returned data being stored in Metaview.
For a custom connector, also review the server’s tools.

​
Access and workspace sharing

Connections are personal by default. Each person sets up their own, and nobody else in the workspace can use it unless an admin shares it. Some workspaces allow only admins to set up connections; other members then use the connections an admin shares. A Metaview workspace admin can share their own connection from that connector’s settings: select Workspace on the connection’s Personal / Workspace control, or Personal to stop sharing. The control is shown only to admins who own the connection. Workspace members can then select the connection from the Connectors menu in Sourcing.
A shared connection uses the connected account’s permissions, so workspace members may be able to read or change information they can’t access through their own accounts. Share only a connection whose data and actions are appropriate for your whole workspace.
Only the connection owner can reconnect or disconnect it, using the Reconnect and Disconnect buttons in the connector’s settings.

​
Disconnect a connector

Open the connector’s settings under Settings > Integrations and select Disconnect on your connection. This stops further requests through it, including for workspace members using it as a shared connection. It doesn’t delete information already returned in a search.