Skip to main content
Once a service is in the Connector Vault, anything that runs in Corvic can use it by name: an agent’s execute_python step, a workflow step, a live app, or a playbook. Corvic’s proxy adds the credential to each request, so the code never handles the raw secret.

​
Give a run access

A run can use a connector only when it has been given explicit access to it.
1

Connect the service

Add it in Settings → Connector Vault. Make it an Organization connector if a shared room needs it; a Private connector is visible only to you and can’t be used in organization rooms.
2

Allow it for the agent or room

Grant the agent access to the connector, or let the room’s policy allow it. Playbooks list the connectors they need, and prompt you to connect any that are missing.
3

Call it from code

Make an ordinary HTTPS request to the connector’s host. Corvic recognizes the host and adds the stored header or query parameter.
MCP connectors appear to the agent as tools. Database connectors give read-only SQL access to the tables the database user can read. Cloud storage connectors let agents list, read, and watch objects in the buckets you named.

​
Pull data into a table

To bring a service’s data into a room as a table instead of calling it from code, use Fetch from an API or a data connector on the Data Sources page. To keep a live copy of database tables, use change data capture.

Agents

Configure an agent’s tools and connectors.

Python & SQL

Run code that calls connected APIs.

Playbooks

Multi-step runs that use connectors.

Access and security

Who can use and edit a connector.