Background space scans
Scans current pages in one Confluence space at a time and keeps progress available while the worker continues in the background.
Atlassian Apps / Confluence
AI Readiness & Content Hygiene Inspector for Confluence
AI-assisted workflows are only as reliable as the content they read. This app helps Confluence space admins scan one space at a time, flag likely hygiene risks under the current policy, and run a review workflow before those pages are relied on more heavily.
- 100% Forge-based
- No external network egress
- Forge hosted storage only
- App-context background scans
Public Material
All the linked review material in one place.
Marketplace buyers, procurement reviewers, and admins usually need more than a feature summary. These links keep the trust, support, and legal material close to the app page.
Why This Matters
Garbage in, garbage out.
If AI-assisted search, summaries, or agent workflows rely on obsolete, incomplete, or conflicting content, the answers become less trustworthy. The app helps teams surface those hygiene risks in a repeatable way.
Erosion of trust in AI
If AI consistently provides inaccurate or outdated answers derived from flawed data, teams will quickly lose confidence.
Poor decision-making
Relying on AI outputs built from obsolete policies or temporary drafts can lead to operational errors and compliance risks.
Manual remediation is unsustainable
For active Confluence environments, manual audit and cleanup of content is a perpetual, unmanageable task.
What The App Does
Space-scoped hygiene review for Confluence pages.
Configurable hygiene policies
Admins can use the shipped presets or tune stale days, minimum text length, score threshold, and detection rules for the space.
Explainable findings
Flagged pages include a score, severity, top signals, and a recommendation so the reviewer can see why the page was surfaced.
Clear review states
The app separates `Needs review`, `Ignored`, `Approved`, and `No issues detected` so admins can tell unflagged pages apart from explicit review decisions.
Space-scoped safeguards
Bootstrap and remediation actions stay tied to the current space, and label actions validate that the target page belongs to that space.
Governance label workflow
The app can apply or remove rovo-ignore as part of the review flow if your team uses that label in its own governance process.
How It Works
Simple for the admin. Structured behind it.
The workflow stays short: tune the policy, start the scan, review the findings, and decide which pages should remain in normal review scope.
- Open it from Confluence Space Settings
Start in the space where the review is needed.
- Set the review policy
Choose a preset or tune stale days, minimum text, score thresholds, and detection rules.
- Start a background scan
The scan continues in the background and persists its progress.
- Review findings by status
Work through open, ignored, approved, or all findings with the signal summary visible.
- Apply governance where needed
Use
rovo-ignoreonly if your team wants that label as part of its own review convention.
Security And Privacy
Trust starts with a narrow architecture.
This page can stay specific because the app's architecture is specific. The app runs on Forge, stays inside Atlassian-managed infrastructure, and avoids external data egress.
Runs entirely on Atlassian Forge
The app does not depend on external servers or vendor-run infrastructure to do its job.
No external network egress
Confluence data is not sent to third-party AI services, analytics, or outside processors.
Forge hosted storage only
App state is stored in Forge storage rather than in a separate external database. Data residency behavior follows Forge platform capabilities and Atlassian controls.
User-context access
Interactive requests run as the current user and remain bound to Confluence permissions.
App-user background execution
Queued scan continuation runs as the Forge app user, and Confluence permissions and restrictions still apply.
Minimal documented scopes
The app requests scopes for app storage, space reads, page reads, and Confluence content writes used for label changes.
Data Handling And Retention
Specific about what is read and kept.
The app reads only what it needs to score content hygiene and stores the workflow state needed to continue scans and review findings.
-
What the app reads
Confluence page content, page metadata, labels, and space metadata needed to score stale, thin, placeholder-like, or otherwise low-trust content.
-
What the app stores
Scan metadata, checkpoints, policy settings, finding records, statuses, and per-page decisions.
-
What it does not keep
The app stores derived workflow data, not a full external copy of the space, and it does not persist raw page bodies in Forge storage.
-
User-linked identifiers
The app does not persist Atlassian account identifiers in scan or decision workflow records.
-
Retention
Scan artifacts and findings use Forge TTL expiry. The current implementation keeps them for 30 days.
Performance And Scale
Built for real spaces.
The app does the heavier work in background batches so the review flow stays responsive and the scan can advance without a fragile long-running browser session.
Queue-backed background scans
Longer scans continue in the background instead of blocking on an open browser tab.
Checkpointed progress
Progress and batch state are persisted so work can resume safely across async runs.
Review stays responsive
Admins review findings separately from scan execution rather than waiting on one long pass.
Designed for more than a tiny test space
The scan works through spaces in batches and keeps findings and status available as the review continues.
Important Product Notes
Clear where the boundaries are.
-
Space admin only
Only Confluence space administrators can use the app.
-
Active subscription required
The app checks for an active Marketplace subscription when licensing data is present.
-
rovo-ignoreis not a native Rovo exclusion controlTreat it as a governance convention for custom instructions and review workflows, not as a documented Atlassian-native feature.
-
Actions stay in the current space
The app validates that label actions target pages in the current space before applying them.
-
Audit visibility depends on Atlassian behavior
Do not assume every action will appear in native Confluence audit logs in the same way.
Pricing
Clear, per-seat pricing.
The pricing scales with your Confluence user tier. These rates are indicative Marketplace rates per user, per month.
| Confluence users | Price |
|---|---|
| Up to 10 users | Free |
| 11-100 users | USD 2.99 |
| 101-250 users | USD 1.79 |
| 251-1,000 users | USD 0.99 |
| 1,001-2,500 users | USD 0.69 |
| 2,501-5,000 users | USD 0.39 |
| 5,001-7,500 users | USD 0.19 |
| 7,501-10,000 users | USD 0.11 |
| 10,001-15,000 users | USD 0.08 |
| 15,001-20,000 users | USD 0.07 |
| 20,001-25,000 users | USD 0.06 |
| 25,001-30,000 users | USD 0.05 |
| 30,001-35,000 users | USD 0.04 |
| 35,001-40,000 users | USD 0.03 |
| 40,001-45,000 users | USD 0.02 |
| 45,000+ users | USD 0.01 |
FAQ
Questions a careful buyer will ask.
-
Does the app send Confluence content to external AI providers?
No. The current architecture avoids external network egress and does not intentionally send Confluence data to third-party AI services or external infrastructure.
-
Where is app data stored?
In Forge hosted storage only.
-
What exactly is retained?
Derived finding data, workflow state, checkpoints, policy settings, and decision state needed to continue scans and support review workflows.
-
Does
No issues detectedmean a page was approved?No. It means the page was not flagged under the current policy.
Approvedis a separate explicit review state. -
How long is scan data kept?
Scan artifacts and findings use TTL-based expiry. The current implementation keeps them for 30 days.
-
Does the app bypass Confluence permissions?
No. Interactive requests remain user-bound, and background scans run as the app user within the Confluence permissions and restrictions granted to that app user.
-
Can a scan keep running after the screen is closed?
Yes. The scan is queued, processed in the background, and resumes from persisted progress.
-
Is
rovo-ignorea native Atlassian exclusion mechanism?No. It is a governance label the app can manage as part of the review workflow, not a documented Atlassian-native exclusion mechanism.
Need A Direct Answer Before Rollout?
Ask it directly.
Support questions, procurement questions, or rollout questions. The path stays direct.