add_connection walks the authorization loop for a new one.
list_mcp_connections
Read-only
List the MCP data sources in a workspace: the ones YOU have connected (status ‘connected’, usable now) plus servers other members registered that you have not authenticated yet (status ‘not_connected’) or that need re-authorization (‘reauth_required’ / ‘revoked’). Use a connected mcpId with list_connection_tools to see what it offers, then with set_widget_pipeline. A not_connected / degraded server cannot be queried or wired until you authenticate it — call add_connection for it.
add_connection
Additive write · Reaches outside Vistazo
Register and/or authenticate an MCP data source for the current user. Call this when a tool returns CONNECTION_NOT_AUTHENTICATED, or when the user names a data source that list_mcp_connections shows as not_connected. Pass the mcpId of an existing workspace server, or a new mcpId + serverUrl (+ displayName) to register a server the workspace has not used before — the record is created for the whole workspace and its URL is validated (public https only in production). Authentication itself happens in the browser: the result carries a connectUrl that opens Vistazo Settings → Connections with that server selected so the user can authorize it. Show the user that link, then WAIT for them to confirm they finished — do not retry the original tool until they say so; OAuth completes out of band.
list_connection_tools
Read-only · Reaches outside Vistazo
List the tools a connected MCP data source exposes — name, description, input schema, and a risk level (safe | caution | destructive). Use this to choose a tool and build its arguments when configuring a widget. “safe” tools are read-only and fine to refresh automatically; “caution”/“destructive” tools should not be wired to auto-refreshing widgets. To try a safe tool once before wiring it (resolve an id, inspect the response shape), use call_connection_tool.
call_connection_tool
Read-only · Reaches outside Vistazo
Run ONE tool on a connected MCP data source directly and return its result. The target is a data source THIS WORKSPACE registered and the caller authenticated (list_mcp_connections lists them, list_connection_tools enumerates what each one offers) — never an arbitrary URL. The discovery lane for building pipelines: resolve ids the schemas tell you not to guess (search/lookup tools), check coverage, or fetch a small sample to see the exact response shape before authoring a transform. Only tools classified ‘safe’ (read-only) may be called; caution/destructive tools are refused. Persists nothing — no widget, dashboard, or pipeline is touched — and results are size-capped, so keep sample fetches small (use the tool’s own limit/filter arguments). The result is unwrapped exactly as a pipeline transform would receive it.