Admin guideSecurity
Scan targets and security.txt
Give a security team the list of app addresses to scan, and publish where to report findings.
Toft publishes what to scan and how to contact you. It does not scan.
Scan targets
Section titled “Scan targets”Admin → Scan targets (GET /api/v1/admin/estate/targets) lists every app with
the URL a scanner should hit — the apps origin on a split-origin install.
- No app is omitted. Apps with no URL (never deployed, batch apps) are still
listed;
apps_addressablecounts the ones that have one. Locked apps are marked. - Visibility is the effective one. A public app reads
authenticated, withvisibility_withheld, when anonymous access is off or reachability has suspended it.
| Caller | Access |
|---|---|
| Administrator | Yes |
| The scanner account | Yes |
Deploy key (toftd_) |
Yes — the only /admin route a key reaches |
| Other signed-in users | 404 |
Every row carries caller_can_reach: whether the account asking could open that
app. apps_reachable_by_caller is the count — compare it with apps_total to see
what your scan will miss.
The scanner account
Section titled “The scanner account”Scan as an account of its own, issued by your identity provider like any other, so your provider owns its lifecycle.
- Sign the scanner in once, so the account exists in Toft.
- On People, open the account and switch on This is a scanner account. Only an administrator can, and not on their own account.
- Grant it on any private apps the scan should cover. The mark gives no access: the account reaches exactly the apps it is granted, like anyone.
- Point the tool at Scan targets with the account’s session or its
personal key, and send
X-Toft-Scan: 1on every request.
A locked account’s mark has no effect.
Marking scan traffic
Section titled “Marking scan traffic”Send X-Toft-Scan: 1 from the scanner account and the request is not counted as
usage, so a sweep does not keep
quiet apps looking busy. It grants nothing else, is ignored
from anyone else, and is stripped before the request reaches the app.
security.txt
Section titled “security.txt”Set on Admin → Settings. Filling in the contact publishes
/.well-known/security.txt on both origins; clearing it removes the file (404).
| Setting | |
|---|---|
security_txt_contact |
Where to send reports. Use a role address, not a person’s. |
security_txt_expires |
Required with a contact. A past date is refused at save. |
security_txt_policy |
Optional https:// page describing how to report. |
security_txt_languages |
Optional language tags the mailbox reads. |
Toft never extends the expiry for you. Admin → System checks warns 30 days before it lapses and again once it has.