Changelog
Spotlight updates automatically through Atlassian Forge - there is nothing to install or upgrade on your side.
Released 11 September 2026
Applies automatically. Nothing for your admin to approve.
Improved
Section titled “Improved”- Updated to the latest Atlassian platform libraries.
Released 11 September 2026
Applies automatically. Nothing for your admin to approve.
Improved
Section titled “Improved”- Publishing to the Team gallery now also needs View. If an admin restricts who can view the gallery, only people who can see it can add to it. Admins who already restrict View do not need to change anything. See Settings.
- Authors can always remove their own shared spotlights, even after an admin changes who can view the gallery.
- More hardening from our ongoing security review. No action needed on your side.
Released 3 September 2026
Applies automatically. Nothing for your admin to approve.
Improved
Section titled “Improved”- Usage analytics now identify your site through Atlassian itself rather than through the browser, so the figures we use to decide what to build next are accurate and cannot be mislabelled. What is sent is unchanged, and still excludes your Jira data, your descriptions and your prompts.
- Hardening across the Team gallery, the dashboard gadget and shared spotlights, from a full security review of the app. No action needed on your side.
Released 3 September 2026
Your Jira admin is asked to approve one new permission before this version applies.
- Usage analytics. Spotlight now records which features get used so we know what to improve. Your Jira data, descriptions and prompts are never included, and admins can turn it off in Atlassian Administration. See Settings.
Improved
Section titled “Improved”- Shared AI plans now wait for a Run plan click instead of running as soon as the Team gallery opens, so a busy gallery stays quick to open. Your own spotlights still run automatically.
- Reports say when they are based on a partial sample, so a partial figure is never shown as a complete one.
- Access tables say when a project has more roles than are listed.
- Averages and minimums counted an empty Jira field as a zero. Figures such as average story points were wrong wherever the field was unset.
- Charts could count the same issue twice.
- AI drafting keeps working when Atlassian retires an underlying model.
1.0.0 - Marketplace launch
Section titled “1.0.0 - Marketplace launch”Release date: TBD
The first public release on the Atlassian Marketplace.
Features
Section titled “Features”- Draft with AI - describe a chart or report in plain language; Atlassian-hosted AI fills the builder with a reviewable draft. Nothing saves without your review.
- Four data sources - issue charts (any JQL; bar/pie/line; seven groupings; four metrics), sprint velocity (committed vs completed per closed sprint), project access reviews (roles and members per project), and AI-composed custom plans (guarded read-only pipelines over Jira, Jira Software, and JSM REST APIs).
- Ten quick-start templates, each showing the AI prompt that would have produced it.
- Spotlights gallery - save, edit, and delete your spotlights; every card keeps the original description.
- Team spotlights - optional site-wide sharing that publishes the definition, never the data; every viewer renders live with their own permissions.
- Group-based governance - a Publish / View permission matrix over Jira groups on the admin Settings page.
- Spotlight Chart dashboard gadget with the same builder.
Privacy & security
Section titled “Privacy & security”- Runs entirely inside Atlassian Forge - no external services, no data egress.
- No write scopes. All Jira access is read-only; custom plans are additionally sandboxed to GET requests on whitelisted Atlassian REST endpoints with hard work budgets.
- Every render fetches live data with the viewer’s own Jira permissions - sharing can never bypass Jira access control.
- Saved spotlights and settings persist in Forge storage (KVS) within your Atlassian tenant; issue data is never stored.
- AI drafting uses Atlassian-hosted models (Forge LLMs) - prompts never leave the Atlassian platform, and no issue data is sent to the model.
- GDPR compliant. Full details on our privacy and security pages.
Known limits
Section titled “Known limits”- Issue charts aggregate up to 1,000 matching issues per render.
- Custom plans are budgeted: max 8 steps, 30 HTTP calls per run, 6 output widgets.
- A spotlight’s visibility can’t be changed after saving (delete and recreate).
- Chart groupings are a fixed set of seven fields - site-specific custom fields aren’t groupable yet.
- Team gallery caps: 200 shared spotlights per site, 15 per author.
- AI drafting fair use: 150 drafts per user and a shared site budget of 1,000 AI calls per month (a draft may use up to two calls).
Future release notes will be appended to this page.