Skip to content

Connect your first service

Nimbus indexes your data locally — but first it needs permission to read from your cloud services. This page walks through adding your first connector, using Google Drive as the worked example.


For a first connector, Google Drive or GitHub are the easiest starting points. Both have straightforward OAuth flows and produce visible index activity within seconds of authorisation.

Nimbus supports the following services:

Category Services
File storage Google Drive, OneDrive, filesystem (opt-in roots)
Email & calendar Gmail, Outlook
Messaging Slack, Microsoft Teams
Photos Google Photos
Source control GitHub, GitLab, Bitbucket
Project tracking Linear, Jira, Notion, Confluence
CI/CD GitHub Actions, GitLab CI, CircleCI, Jenkins
Incident management PagerDuty
Observability Sentry, Grafana, Datadog, New Relic
Infrastructure Kubernetes, AWS, Azure, GCP, IaC tools (Terraform, Pulumi, CloudFormation)

For per-service setup details, scopes, and known limitations, see the Connectors reference.


In the Nimbus desktop app, click Settings in the sidebar (or press S), then select Connectors.

Click the Google Drive tile. If no connector is configured yet, you can also reach this from the onboarding wizard on first launch.

Click Authenticate. Nimbus opens your system browser to Google’s OAuth consent screen. The scopes requested are read-only by default:

Scope Purpose
drive.readonly List files and folders; read metadata
drive.file (exported text types only) Read document text for indexable formats

Write and delete scopes are never requested unless you explicitly enable a write tool that triggers a HITL-gated action. Nimbus requests the minimum scope set required for the tools you have actually enabled.

Approve the scopes on the Google consent screen. Your browser will show:

“You may close this window.”

Switch back to Nimbus. The Google Drive connector row now shows a green health dot and the status Enabled.

Connectors panel with a connected Google Drive row

Once connected, the initial sync starts immediately. The following metadata is indexed for every file:

  • File path and name
  • MIME type
  • Last-modified timestamp
  • File size
  • Owner display name

Text content is also indexed for indexable types: Markdown, plain text, Google Docs (exported as plain text), Google Sheets (cell values), and Google Slides (speaker notes + slide text). This is what enables semantic search — “find the Q3 planning doc” resolves to the right file even if you don’t remember its exact name.

For Google Photos, only metadata is indexed: album name, capture timestamp, and location (if you opt in to location data in the connector settings).


Not indexed Reason
File permissions and sharing settings Out of scope for the connector surface — Nimbus indexes content and metadata, not access-control lists
Binary file content (images, PDFs, compiled artefacts) Not text-extractable without a conversion step
Trashed / deleted files Not surfaced by the Drive API delta feed
Files in shared drives you do not have read access to Access is validated at OAuth time

Sync type When
Initial sync Runs immediately after a successful OAuth handshake
Incremental sync Every 60 seconds by default
Delta sync Drive, Gmail, and OneDrive use the API’s native change token — only changed items are fetched

You can adjust the per-connector sync interval with nimbus connector set-interval <service> <duration> (for example nimbus connector set-interval google_drive 5m). The desktop app’s Settings → Connectors → (connector name) → Sync interval does the same once the desktop app is released.


Repeat the same flow for each additional service. Multiple connectors sync in parallel; nimbus connector list shows each connector’s item count and last-synced time. There is no limit to the number of connectors you can configure.


To remove a connector, run nimbus connector remove <service> (the desktop app’s Settings → Connectors → (connector name) → Remove does the same once the desktop app is released):

  1. nimbus connector remove google_drive
  2. Nimbus asks you to confirm before deleting the stored OAuth credential (from the system vault) and all indexed items from that service.
  3. Because this is a destructive operation, removal is HITL-gated in the Gateway; the CLI’s confirmation is the answer to that consent request, so you are asked once. Decline and nothing is sent.

The removal cascade deletes the vault credential first, then the indexed items. If you reconnect the same service later, a fresh initial sync re-populates the index.