Confluence
Confluence is where a lot of teams already write decisions down - and where those write-ups go stale without anyone noticing. Align captures new decisions from comments and page publishes, and Discover scans what is already there.
Confluence is fully built and shipped as a container image, but connectors.confluence.enabled
is false in the chart's default values. On Align Cloud, ask us to enable it. Self-hosting, set
the flag - see Configuration.
What it does
- Capture from comments -
/alignin a page comment - Page-publish signal - publishing a decision-shaped page can raise a capture suggestion
- Discover - scan existing spaces, decision logs and RFC pages for past decisions
- Evidence linking - a decision keeps a link back to the page and comment it came from
Setup
- Settings → Integrations → Connect next to Confluence
- Sign in with Atlassian and select the site to grant access to
- Choose which spaces are in scope
Self-hosted deployments use Align's centrally-managed Atlassian OAuth app - you do not need to register your own. See Connector Authentication.
Capturing a decision
In a comment on any page in an opted-in space:
/align We're standardising on OpenAPI 3.1 for all public service contracts
Align reads the surrounding page context, synthesises a Decision Snapshot, and replies with a link to it.
What Confluence cannot do
Two limits come from Atlassian's API, not from Align, and it is worth knowing which is which:
- No react-to-capture. Atlassian Cloud emits no comment-reaction webhook events, so the
📌 / 🔖 reaction path that works in Slack and Teams is not available here. Use
/align. - Page publish bypasses per-space opt-in. A publish is already an explicit, deliberate act by an author, so it is treated like an explicit trigger rather than an ambient signal.
OAuth scopes
| Scope | Why |
|---|---|
read:page:confluence | Read page content for context |
read:space:confluence | Enumerate spaces you have granted |
read:comment:confluence | Read the comment that triggered a capture |
read:confluence-content.all | Discover scans across granted content |
offline_access | Refresh the token without re-authorising |
These are read-only. Align does not write to your Confluence content.
Jira and Confluence use the same Atlassian OAuth application. Connecting one and then the other reuses the same authorisation, and a refreshed token keeps the scopes it was originally granted - so if you connected Jira before Confluence support existed, you may need to reconnect to pick up the Confluence scopes.
Evidence captured
- Page link and title
- Space key and name
- Comment link, where the capture came from a comment
- Author and timestamp