Confluence

Confluence context enriches an existing Pulse. It does not create Pulses.

How this helps triage: matching excerpts can show up in the summary, the first checks, and What informed this on the Pulse.

Exhale pulls Confluence pages by label or space when a Pulse matches a service key, then attaches the top-scoring heading sections as runbook context.

Connect

1. Sign in as a workspace admin.
2. Open Connections → Context → Runbooks & docs → Confluence.
3. Authorize Confluence OAuth (Atlassian) and choose spaces or labels. Default scopes are read:confluence-content.all, read:confluence-space.summary, search:confluence, and offline_access. Reconnect with write access also requests write:confluence-content.
4. Test connection to mark the connector configured.
5. Disconnect clears the stored token.

Confluence access tokens last about an hour; Exhale refreshes them with the stored refresh token before listing pages or reading content, and asks you to reconnect if that refresh fails.

Missing credentials, timeouts, or vendor 403s skip this block. Triage still completes. Send-test-alert Pulses skip live collectors.

Plan

Growth or higher.

Connector slots are shared across Context categories (markdown runbooks use a separate document cap). Tighten AI → Prompt → Context use if excerpts are noisy or too large.

API

Canonical paths are under /v1. Writes use the signed-in session and CSRF. Secrets are never returned on GET. Disconnect is DELETE /v1/auth/context-connectors/confluence. Start is GET /v1/auth/context-connectors/confluence/oauth/start.

On Pulse Detail postmortems, Growth or higher workspaces can also create a Confluence page from the scrubbed summary when this connector is configured.

Related

• Context index
• Prompt → Context use