API: SSL endpoints, key rotation, repository passphrase, server upgrade
API: document storage location updates/deletes and remote SSH target management
Drop references to an unannounced product from the public API page
The Session API section said these endpoints back a mobile app, and the SSO
note referred to "the app" as a known thing. Neither is ours to announce here.
Both now describe the capability instead: anything that can hold a bearer
token.
Document the API additions through v2.91.1
Audited every /api/v1 route in App.php against this page, then every field
added to endpoints it already listed — which is where the gaps were, since a
new key on an existing endpoint leaves the route list looking complete.
Four were missing:
- GET /api/v1/dashboard gained archives and jobs_24h. Both are written down
with the traps: on_disk_bytes is repository size rather than summed archive
dedup and the two drift; dedup_savings_percent is clamped server-side so
every client rounds the same way; jobs_24h always returns 24 buckets; and its
failed series counts failed jobs, which is deliberately not the same measure
as the web chart's red bars (#240) — noted so nobody files the mismatch as a
bug.
- GET /api/v1/server-stats is new, admin-only, and returns numbers rather than
the formatted strings the web equivalent produces for its own JavaScript.
Null network figures mean unreadable, not idle.
- The health backups check changed meaning, not just shape: failed_24h no
longer moves the status. Anything alerting off that number now disagrees with
status, so it says so plainly.
- A client profile's schedule carries a timezone, and applying one writes it.
/api/v1/platform/rotate-token stays undocumented on purpose: it is gated on the
hosted platform token, so it is control-plane machinery rather than something a
customer can call.
Document the per-profile backup-overdue threshold (#409)
Explains why retries cannot stand in for it, and that failures inside a
client's allowance are reported but do not change health status.
API: document the push service and client profile endpoints
Also brings the GET /settings sample back in line with what the server
returns — it was missing the agent settings and the Apprise keys added since
it was written — and records the fourth rule that now holds: values are
effective, so a setting never explicitly saved reads as the default actually
in use rather than 0/false.
API reference: neutral example in the sign-in response too
API reference: neutral example in the sign-in request
The login example used a real name and a device brand. Neither adds
anything to the documentation, and both say more about who wrote it than
the endpoint does.
API reference: client install and stats, repository maintenance, restore
Brings the page level with the server again — 16 endpoints had shipped
undocumented.
Clients: the install command (and why it is the one place an agent key is
returned), and the stats endpoint with the windows it averages over.
Repositories: the six maintenance actions, catalog rebuild, recovery-point
deletion, and what a restore point holds — including has_databases on
archive listings so only usable restore points need be offered.
Restore: catalog search, download with its 507 space response, database
connectors and the single restore endpoint.
Also health (both the unauthenticated probe and the detailed check, with
what raises each state), plugin configuration writes and their write-only
secret handling, registered push devices, catalog freshness on the file
browse, and the jobs status filter now taking a list.
Every /api/v1 route is documented except the hosted platform's own
token rotation, which is intentionally not customer-facing.
API reference: document the full /api/v1 surface
Brings the page up to date with everything added since it was written —
89 endpoints were live and roughly half were undocumented.
- Two token kinds (admin vs session) and how scoping differs between them
- Job detail by job id, with logs, queue position, current file and prune
stats; queue history, slots and 24h stats; cancel and retry
- Schedules day view, including how occurrences are placed in the
viewer's timezone and what each state means
- Dashboard, notifications, catalog browse, restore
- Session API: discover, login, 2FA, brokered SSO, sessions, push
- My Profile: account, timezones, password, storage alerts, 2FA, reports
- Server settings, notification services, backup templates, API tokens,
updates (all marked unreleased)
- Monitoring: metrics snapshot and fleet summary
- Expanded the error table (403/404/409/410/422 now mean more than they did)
Endpoints are labelled with the release that introduced them; the
settings group is marked unreleased until it ships.
Add comprehensive UI screenshots throughout the wiki
- Capture 26 screenshots of a populated demo deployment (dashboard, clients
and all client tabs, queue, schedules, log, storage, settings tabs, users,
notifications, repository/job detail, create-repo/plan forms, 2FA, SSO, API).
- Replace ~90 'Screenshot:' placeholders with real images; remove placeholders
that had no corresponding screen.
- Add hero images to Home, Dashboard, Repositories, Notifications, Remote
Storage, Storage Setup, SSO, and API pages.
- Fix broken internal links ([[Agent Setup]]/[[Agent-Setup]] -> Linux Agent
Setup, [[Restore]] -> Restoring Files, backwards Standard Installation anchor).
API docs: recovery point lock/unlock endpoints and delete guards (#314)
API docs: users, server log, and schedules endpoints
API docs: document POST and DELETE /s3-credentials
API docs: clarify that non-secret responses work without the capability
API docs: s3-credentials now uses include_secrets flag for consistency
API docs: document Display Secrets token capability
API docs: GET /api/v1/repositories with optional include_secrets
API docs: add GET /api/v1/s3-credentials
API docs: add maintenance-mode GET/POST endpoints
API docs: storage create + capacity, per-repo S3 toggle, repo response fields (v2.55.0)
Add edit/delete/trigger/jobs/queue docs to API reference