StackJack 0.1.0.13529+8ee5e80
StackJack 0.1.0.13529+8ee5e80
This release adds several AI Governance capabilities: exposure scans for clients and prospects, agent rules with approvals, one-click fixes, incident rules, and a page for applications and automations. Many long AI Governance pages are now split into tabs you can link to, and the endpoint agent's device list is now a searchable table.
New features
Exposure scans for client organizations and prospects. Run a scan from an organization's page to see which apps have been granted Microsoft 365 permissions and which AI vendor accounts and keys exist.
Vendor coverage includes Anthropic and OpenAI admin keys, GitHub Copilot, and Google Workspace.
Each finding is rated High, Medium, or Info, and the scan lists what to fix first.
You can compare two scans to see what was added, removed, or changed.
You can download a PDF report. The report identifies vendor keys by vendor and type, not by person name.
You can add a vendor key that is used only for the scan. Removing that key from the scan also removes the stored key.
Prospects. You can add prospective clients as their own kind of organization, separate from your client list, and run an exposure scan for them.
For Microsoft consent, you can generate a single-use link that expires after 72 hours and send it to the prospect's administrator.
Prospects that are not converted are closed automatically after 30 days.
Agent rules. A new Agent rules page lets administrators control which tools your AI agents can call.
Rules can run in Test mode (observe only) or Live mode.
In Live mode, a rule can refuse a call or hold it for approval.
Held calls appear on an approval page and in Microsoft Teams approval cards.
An approval is used once, for the call that was approved.
A daily alert reports how many calls your rules matched.
Apply fix and Undo. From a finding or alert, a new Act menu lets you:
Block or unblock a domain in DNSFilter, Cloudflare, or Cisco Umbrella. "Undo this block" removes the block from the list that block used, including older blocks.
Open a PSA ticket.
Ask for a decision on an inventory item.
Fix settings. A new tab where you choose where blocks are applied.
Incident rules. Rules can group matching findings, alerts, and activity into incidents automatically.
Each rule has a threshold.
Each rule can either add matches to an open incident or open a new incident each time.
A daily cap limits how many incidents are opened; matches beyond the cap go into a single overflow incident.
You can preview a rule before turning it on. Incident rules have their own tab under Alerts.
Applications and automations. A new page lets you record non-person users of AI, such as scripts, services, and integrations, so their activity is no longer treated as an unknown person. You can:
Create applications and set what activity they match.
Apply them to past activity.
Convert or merge detected sign-ins into an application.
Attribute a single record to an application.
Attribute this key on the spend page. The spend page lists the vendor API keys seen in the selected period, with requests, tokens, the vendor-reported cost, and who currently holds each key.
Administrators can assign a key and its past usage to an existing or new application or automation.
Moving a key that is already assigned asks for a second confirmation that names who loses it.
Reviewers and auditors can view the list but cannot change it.
Text reveal sessions. Instead of a separate reason for every reveal, you can start one session with a reason and a length.
A chip shows that a session is active, and signing out ends it.
Each reveal is still recorded individually in the audit trail.
Each organization's maximum session length is applied to that organization's text throughout the session.
Actionable alerts. Each alert now:
Includes guidance and says what it is about.
Names the tool involved on the row.
Offers direct actions: approve, deny, or open an incident.
You can also acknowledge or resolve alerts in bulk.
Activity drawer actions. From an activity record you can:
Attribute it to a person or application.
Open an incident, or add it to an open incident.
See related activity.
Findings explain what would happen in Live mode and offer "Make this rule live". The Activity page has a new session filter.
New tools for your AI agents. Agents can now view and compare exposure scans and propose new ones, list agent rules, and open tickets or request decisions. All of these still respect your roles and approval settings.
Improvements
- Endpoint agent device reporting. The endpoint agent now reports the computer's name, Windows version, and when it last read its policy. Older agents continue to work.
- Endpoint agent install options. The install card now presents two separate ways to install on each platform: the MSI installer or the script.
- The MSI option includes the silent install command, a table of MSI settings, and the silent removal command.
- The card also includes a note on deploying with Intune.
- Endpoint agent icon. The StackJack-branded endpoint agent now shows the StackJack icon, including in Windows Apps and features, where it was previously blank.
- Billing page. Applications and automations are now shown on the Billing page and in the plan description alongside staff seats.
UI & UX changes
Tabs on long pages. Many long AI Governance pages are now split into tabs. Each tab has its own address, so links, the Back button, and refresh return you to the same tab. The tabs support keyboard navigation and show count badges where relevant. Pages with new tabs:
Billing
Alert settings
Privacy and notices
The organization page
Managed settings
Staff
Staff sign-ins
Cloud AI
Microsoft 365
Browser extension
Report packs
Client names
AI judgments
Endpoint agent page. The page now has five tabs: Overview, Install, Policy, Devices, and Branding. Branding was previously a small link at the bottom of the page, and its old address now opens the Branding tab.
Endpoint agent device table. The Devices tab shows each device with:
Computer name, person, and Windows version
Agent version, with a chip when an update is available
Platform, enrollment date, last seen, last policy read, and health
You can filter using summary cards, sort, search, and page through results. Opening a device shows its last report in a side panel. You can revoke or forget devices in bulk, with a typed confirmation for each action.
Finding review. Assigning an Open finding now moves it to In review. The separate Start review button has been removed.
Application names. Alerts, report packs, the data-request log, and the export list now show an application's own name instead of "Erased person".
Teams approval cards. When an approved call proceeds, the Teams card now reads "Approved and used", matching the Portal.