Integrations setup

Every connection is made by a workspace admin, scoped to your workspace, and revocable. Here’s how each one is set up and what it unlocks.

  1. 1 During onboarding (or later from Settings → GitHub), pick the personal account or organization to connect and click “Install app”.
  2. 2 On GitHub, choose all repositories or a selection — that’s the boundary of what Gentlybot can see. You can change or revoke it on GitHub at any time.
  3. 3 Optionally enable autonomous capabilities in settings: auto-fix failing CI, address review comments, PR-comment slash commands (@gentlybot …), and security-alert fixes. Each is a separate toggle, off by default, with its own scoped GitHub permission.

Gentlybot clones code on demand into sandboxes; it does not keep a standing copy of your repositories.

  1. 1 A workspace admin or owner connects Slack from Settings → Slack; you’ll go through Slack’s standard authorization screen.
  2. 2 Invite @gentlybot to the channels where work happens.
  3. 3 Mention it to start a run, attach screenshots directly, and get updates and results posted back to the channel.
  1. 1 Connect from Settings → Integrations with Atlassian OAuth and pick your site.
  2. 2 Threads can then create well-scoped issues (with the PR and evidence linked) and read issue context into runs.

Automation triggers on Jira events are in early access.

  1. 1 Connect from Settings → Integrations with Notion OAuth and grant access to the pages you want available.
  2. 2 Runs can read those pages as product context.

Early access.

  1. 1 From Settings → MCP, add a remote MCP server by URL. Authenticate with a bearer token or OAuth, depending on what the server supports.
  2. 2 Pick from presets for common tools (Confluence, Linear, Sentry, Datadog, …) or add any server that speaks the Model Context Protocol.
  3. 3 The connection panel shows live health, so a degraded integration is visible instead of silently failing mid-run.

Credentials are encrypted at rest and scoped to your workspace.

  1. 1 Capable models are included out of the box — nothing to configure to get started.
  2. 2 Admins can adjust model pools in Settings → AI: which models handle asking, coding, and other kinds of work, with fallbacks.
  3. 3 Enterprise workspaces can bring their own model accounts or private deployments.