Compliance answers
Settings, Compliance (/admin/compliance), gated by the user.manage
permission.
This screen answers the questions on a departmental vendor assessment from your deployment’s own configuration, so a school filling in the form does not have to email us and ask. Each section loads on its own: one that cannot load says so, and is never shown as compliant.
Where your data lives
Section titled “Where your data lives”Hosting (managed by Plugboard, or self-hosted), Location and Residency, and on managed hosting the Region id. Read from local configuration rather than by calling home, so it still answers while cut off from everything, including during an incident. Backups and connected services can process data elsewhere, and the residency line says so. See regions and data residency.
What reaches Plugboard
Section titled “What reaches Plugboard”Support access
Section titled “Support access”On managed hosting, Plugboard support has a channel into your deployment. It is Off until an administrator here grants it, for a set time and with a reason. While it is on, support can read deployment and administrator details, list backups, and email an administrator a link to set their password. It can restore a backup only when the grant includes restores.
To grant it, someone holding user.manage fills in:
| Field | Choices |
|---|---|
| How long | 24 hours, 3 days (selected) or 7 days |
| Reason | 10 to 500 characters. Support and your audit log both see it |
| Restoring backups | Include restoring backups, unticked unless you tick it |
Allow support access asks you to confirm it is you. The row then reads On until the expiry, with who granted it and why, and whether restoring backups is included. Granting again while access is on replaces the grant, so it extends or shortens it. End access now ends it straight away, without a confirmation, and works while the desk is read-only.
Expiry needs nothing to run: once the time passes, support’s next request is refused.
Support activity lists the last 20 support entries, each with who and when, and Open in the audit log shows them there. Every support request is recorded in your audit log before it reads anything, and so is every refused one.
A setup link goes only to the address on an administrator’s account, through your school’s own outgoing mail, and never to support. Without outgoing mail the request is refused.
A school with a single administrator is warned: if that person loses access, nobody can grant support access, and recovery needs a written request from the principal or business manager.
Support access governs the product’s support channel. It does not change hosting administrator access to the server we run for you, and the page says so. A deployment with no Plugboard support key, such as a self-hosted one, shows None: any server or remote access is yours to grant separately.
Bug reports
Section titled “Bug reports”Separate from support access, and pointing the other way: this sends your words and your name outwards. With Let staff send reports to Plugboard on, staff can report a fault or ask for something from a button in the staff header. A report goes to the people who build Plugboard, not to your own service desk, with the reporter’s name and email address so we can answer them.
On managed hosting it is on by default, because we already run the deployment for you. On a deployment you run yourself it is off by default, and nothing is sent unless somebody turns it on. Switch it off and the button disappears and nothing leaves; nothing else about the product changes. A deployment with no licensed route to Plugboard shows None, and staff report problems to us by email.
Sent by your staff lists each report with who sent it and where it got to: Not sent yet, Could not be sent, Received, Planned, Being worked on, Fixed and released or Not planned, with our reply when there is one.
Who else touches your data
Section titled “Who else touches your data”Every connector configured here, with its category and the kind of data it handles. Stated per category, because the question is about the kind of data leaving, and a school swapping Jamf for Intune has not changed the answer. A connector that reaches a system on your own network is marked private network, and one that is switched off is marked Off.
| Category | Data it handles |
|---|---|
| MDM | Device serial numbers, models, assignment and compliance state |
| SIS | Student and staff names, year levels, identifiers and email addresses |
| Directory | Names, usernames, email addresses and group membership |
| Identity | Usernames and email addresses, for authentication only |
| Message content and recipient addresses for notifications sent by the desk | |
| SMS | Mobile numbers and message content |
| Printing | Usernames, card numbers and print activity |
| Ticketing | Ticket subjects, descriptions and the requester’s name |
| Warranty | Device serial numbers only |
| Repair vendor | Device serial numbers, fault descriptions and the owner’s name |
| Spreadsheet | Whatever the linked sheet contains, which is usually induction rosters |
| CRM | Organisation and contact records |
| Billing | Billing contact and payment records. No student data |
| Security | Device identifiers and policy decisions |
| AI | The text of a question and the desk records needed to answer it. Where it goes depends on which AI connector you configured; see the AI row |
The list describes this deployment only. Review hosting and support access separately.
What one account did
Section titled “What one account did”For an incident assessment, and needs audit.view. Find an account by name or
email; a deleted account the audit log still names is found too, or by the ID in
an audit entry. The report says how many actions it took and when, and which
kinds of record it touched, and Download full activity (CSV) gives the whole
list.
Using it for an assessment
Section titled “Using it for an assessment”Screenshot this page. It answers, in the assessment’s own terms:
- Where is the data held
- Is it transferred outside that jurisdiction
- Which sub-processors receive what
- Can the vendor access our data, and who decides
- What did one account do, for an incident
The parts an assessment asks that this page cannot answer are the ones about your own configuration: who at your school has which permissions, how long you keep the audit log, and whether you have restored a backup. Those are roles, audit and retention and backups.
Erasure and retention
Section titled “Erasure and retention”Two different clocks.
Audit retention is operational and set by you. Longer is better for investigation and worse for disk and for privacy. See audit and retention.
Erasure removes a person’s records on request. It is built into the product rather than a manual database edit.
The AI row
Section titled “The AI row”Where ticket text goes depends on which AI connector you configured, and the three answers differ:
- The bundled model, which runs inside your own deployment. Inference stays on that configured endpoint; enabled connectors, notifications and external clients have their own data flows.
- Your own key, through Anthropic or an OpenAI-compatible endpoint. Ticket text goes to that provider under your agreement with them, or to your own GPU box if that is what you pointed it at.
- Vendor-hosted AI, on managed deployments. Pseudonymised ticket text goes to Cloudflare Workers AI for embeddings and decisions, and through Cloudflare to Anthropic’s Claude, processed in the United States, for chat and drafts. See sub-processors.
The connector list names the one you have. With no AI connector at all, nothing goes anywhere: Ask answers from the desk’s own records, and nothing that needs a model is shown.
None of that extends to a third-party MCP client you choose to connect, because that client hands tool results to whatever model it uses.
In an assessment, state it this way: which AI connector is configured decides where ticket text goes, this page names it, and connecting an external MCP client is a separate decision you make. Scope API keys accordingly.