MB1 (spec 13). Decision 0005: our own Android/iPhone app, Expo in the
mobile repo, user portal only, no push in v1.
- mobile_devices (migration 0013): one-time pairing codes from My settings
(QR or server + code, 10 minutes), traded by the app for an spm_ device
token that acts as its user and never as an admin
- POST /api/me/mobile-devices/pair (web session only), POST /api/mobile/pair,
GET/DELETE /api/me/mobile-devices
- signing out everywhere (password change or reset, disabling) removes
paired phones too
- My settings: Mobile app section with the QR code and paired phones
- mobile/ (its own repo) ignored here
Two separate reviews of the SSO branch and all 207 API routes (spec 12 §8.6). Every route was
guarded as meant. Fixed, each with a test:
- login: a directory 503 counts as a failed sign-in; a locked directory account is a refused
password, not "unavailable"
- open redirect after OpenID sign-in (a tab in next)
- directory email matching only for non-admins without their own password; others connect on
My settings (POST /api/me/identities/ldap); only admins disconnect a directory account
- OpenID session check reads the refreshed ID token's claims too
- connect is POST /api/auth/oidc/connect (CSRF header)
- stored bind password, client secret and SMTP password tied to their server
- 409 role_managed when a provider sets the role; empty Contrack PATCH is a 400
- @fastify/rate-limit on every route (RATE_LIMIT_PER_MINUTE), 10/min on Test buttons; a
password asked again shares the login limiter
Live tests pass against lldap 0.6.3, Samba 4 AD and Keycloak 26.8.1.
Spec 12 S2: openid-client 6.8 (decision 0004). Discovery, PKCE + state + nonce, ID token
signatures checked, userinfo always read. Users connect their OpenID account from My account
and sign in either way; no accounts are made; a user without a password is matched once by
verified email. auth.mode mixed and oidc (admins keep a password way in), role from adminGroups,
allowedGroups, signing out at the provider. Settings -> Sign-in with Test. Tested against a fake
provider with RS256 and ES256 keys.
C1 (spec 11): /api/me/contacts and /api/contacts (admin), our own vCard 2.1/3.0/4.0 and CSV
readers with a preview, contact_shares (migration 0009) for offers, and contact names in call
history, voicemail, Home, the softphone and the call report email. Admin → Contacts.
The server checks itself (public IP, DNS, local networks, phone ACL, SIP transports, free audio
ports) and says how to fix each problem. A quick or complete network test asks the probe
(apps/probe, own image) to test this server from outside: HTTP/HTTPS, SIP over UDP/TCP/TLS, the
browser phone's /ws, sampled audio ports. The probe only tests the address that asks.
The audio port range comes from RTP_PORT_START/END (default 10000-20000).
The 08 §1 threat table checked line by line against the code, plus a review by a separate agent
that hadn't written it. Findings were reproduced with a failing test before each fix:
- Critical: "011…,4" turned the route's Goto into Goto(sp-route-…,011…,4,1), starting past the
class check, so any extension could call abroad. FILTER on the dialed number (also stops "&"
adding a Dial target). trunk-test.ts proves it.
- High: -web endpoints (desk phone's password, any transport) skipped the phone ACL; now
127.0.0.1 only when it's on. call-test.ts: UDP refused, WebSocket from 127.0.0.1 registers.
- IVR direct dial no longer reaches paging groups or parked calls (sp-ivr-direct).
- Trunks identify_by = ip: a From of trunk-<id> from anywhere no longer rings phones.
- .dockerignore is an allowlist, so the public source archive can't pick up stray files.
- CSRF check on the matched route (/%61pi/… got around it); Linphone tokens masked in logs;
/api/live checks Origin and re-checks the session; *45/** filter digits; pg_dump password in
the environment; more webhook ranges blocked.
The review and what's left for the owner are in 08 §6.
Task: M11-T8
System page laid out after 3CX's admin dashboard: a System information panel (version,
host name, company, public IP, LAN IPs, ports, telephony, restart needed, copy buttons,
how to restart) and an Event log panel (system_events, 90 days: trunk down/up, Asterisk
down/up, restart needed, config failed, 911, public IP changed, backups, disk) with Load
more and a CSV download.
About (/about, no sign-in, linked from the login page, My phone and System): AGPL-3.0
notice, credits, the repository link, and "Code v<version>", which downloads
selfphone.tar.gz: the api image now tars its own build context (no .env, data,
node_modules, build output or .git) before installing anything.
Task: M11-T7
GET /api/reports/summary and /api/reports/queues (company-time-zone days, last 30 by
default, a year at most) and the Reports page: totals, stacked bars per day, missed per
day, a weekday x hour heatmap and the queue table. Plain divs, no chart library.
Task: M11-T6
API tokens (Settings -> API tokens): Authorization: Bearer sp_<40 base62>,
stored as sha256, shown once. A token acts as its admin with their current
role and stops when they're disabled; Bearer requests skip the CSRF header
and never fall back to the cookie. A token can't manage tokens.
Webhooks (Settings -> Webhooks): call.started/answered/ended,
voicemail.new, emergency.call, trunk/extension registration changes, from
the live state; queued per webhook, sent every 10 s with
X-SelfPhoneNG-Signature (HMAC of timestamp.body), retried after 1 m, 5 m,
30 m, 2 h, 12 h, then failed. Private/loopback targets are refused in the
connection's own DNS lookup (no rebinding), unless allowed. Send test and
a delivery log per webhook.
The e2e found voicemail.new missing for mailboxes without a user: the
voicemail scan now queues it for every mailbox. vmpin.test.ts gets 180 s:
Git Bash forks make it ~85 s on Windows (the earlier flaky gate run).
Task: M11-T5
Headers (08 §2.4) on every response: CSP (connect-src also allows
WEBRTC_WS_URL's origin when set), nosniff, Referrer-Policy,
Permissions-Policy. The dark-mode bootstrap moved from an inline
<script> to /theme.js so script-src stays 'self'.
Phone ACL (08 §2.2): with sip.restrictPhonesToLocalNets, every desk
extension endpoint gets deny 0.0.0.0/0 and permit for each local and
extra net. The spec's snippet was wrong: contact_acl names acl.conf
sections, a pjsip type=acl is the global ACL (it would filter trunks),
and Contact-header rules let a NATed remote phone through. Browser
endpoints come through the reverse proxy and aren't filtered. Golden
case phone-acl. Settings → Security holds it, with the provisioning
switches.
PIN rules for PINs changed from the phone's voicemail menu, which went
around the api: externpasscheck = sp-vm-pincheck (in the Asterisk
image, same rules as isTrivialPin, a unit test keeps them in step) and
minpassword = 4. Accepted PINs reach the api through AMI
VoicemailPasswordChange and are stored; before, the next config write
silently put the old PIN back.
e2e m11-t1: no CSP violations across admin pages; Asterisk refuses a
REGISTER ("Not match Endpoint ACL") until the network is listed; from
the phone, 1111 is refused by the check script and 7391 is stored and
survives a later config write.
Task: M11-T1
/api/devices CRUD (MAC in any format → 12 lowercase hex; a 20-char
provisioning password, stored encrypted and shown to admins, who type
it into the phone) and GET /:id/config-preview with passwords masked.
GET /prov/... per 07 §2: <mac>.cfg (Yealink), cfg<mac>.xml
(Grandstream), the empty Yealink common file. Basic auth with the MAC
and the device's password (timing-safe), 404 for unknown MACs or the
other vendor's file, 403 outside the local nets when LAN-only, 404 when
provisioning is off, Cache-Control no-store, last provisioned time and
IP recorded. Limits: 20 requests / 10 min per IP outside the local nets
(an office's phones rebooting together would pass it), and 10 failures
block the IP for an hour. A first request without credentials isn't a
failure: phones ask that way and retry after the 401.
Templates (pure functions of a ProvisionContext) hold the account,
voicemail, pickup codes and the provisioning settings; BLF, time zone
and phonebook come with M10-T2.
Web: Phones page (table, drawer with what to type into the phone,
config preview; manual settings for generic phones). e2e m10-t1: phone
added in the UI, 401 then the file, and a baresip phone registers with
only the account from it.
Also: the loadSnapshot test made its database inside the test (20 s);
under a full parallel run that timed out now and then. It now does it
in a hook like the others.
Task: M10-T1
/api/moh-classes: create (slug, not "default"), delete (409 while an
extension or queue uses it), add/play/remove files. Files are 8 kHz
wav in MOH_DIR/<class>/ and sound_files rows of kind moh. Extensions
and queues may only pick default or an existing class.
Two things found with real calls, written back into 03 and 04:
- An empty class plays silence on Asterisk 24. Only classes with
music are rendered, so a channel asking for an empty one gets
default.
- The generator reloads only changed files, and a file upload changes
no config (except a class's first/last). The route sends moh reload
after every file change.
Web: a Music on hold section on Sounds & music, and a select of sets
in the user drawer's Phone tab and the queue form. load-golden makes
the class folders like the api does. Golden case queue gains two
classes. e2e m8-t5: set made in the UI, upload listed by Asterisk, a
queue caller hears it, a user's moh_suggest, delete refused while
used, an emptied set falls back to default.
Task: M8-T5
/api/paging-groups CRUD (members as memberIds). The generator renders
sp-page-{id} per 03 §6.11 (desk phones only, auto-answer headers via
sp-sub-autoanswer) plus the number in sp-local-numbers. Added the `s`
option so phones already on a call aren't paged, which also keeps a
pager who is a member from being paged themselves. A group with no
members just hangs up.
*80<ext> intercom checks the target first: digits only and an endpoint
in sp-from-ext, else ss-noservice (the spec's one-liner passed
${EXTEN:3} straight into Page()). Both written back into 03.
Web: Paging page (list, drawer form with member checkboxes). Golden
case paging. e2e m8-t4 with baresip phones that honor the headers
(sip_autoanswer): an ordinary call rings, a page is answered by itself
and muted, the busy pager is skipped, *80 is two-way, *80999 refused.
The check on a real Yealink/Grandstream stays open (09).
Task: M8-T4
/api/conferences CRUD: PINs stored encrypted, shown to admins (they hand
them out), never in the audit log; an admin PIN needs a different room
PIN (checked against the stored values on PATCH). The generator renders
sp-conf-{id} per 03 §6.11 (PIN prompt, 3 tries, admin PIN joins with
sp_admin_user), the number in sp-local-numbers and destApp conference.
Two changes from the spec's snippet, both found with real calls and
written back into 03:
- Set(CONFBRIDGE(bridge,max_members)) is ignored once ConfBridge()
names a bridge profile: a 4th caller got into a room of 3. Each room
now has its own [sp_room_<n>](default_bridge) profile in
confbridge_selfphoneng.conf.
- A recorded room runs sp-sub-record on each caller's leg (just before
ConfBridge) instead of record_conference, so the files are ordinary
call recordings: call history, Recordings page, retention.
Web: Conference rooms page (list, drawer form), "Conference room" in
the DestinationPicker. Golden case conferences. e2e m8-t3: room built
in the UI, wrong PIN refused and asked again, admin PIN joins with
flags AM, a 4th caller turned away, recordings listed, and a no-PIN
room with two desk phones.
Task: M8-T3
Queues per 03 §6.8: queues_selfphoneng.conf (members as Local agents
following their hint), sp-queue-{id} (answer, prefix, record, Queue with
max wait, timeout destination), the number in sp-local-numbers and
destApp queue. Golden case queue; the Asterisk load test now checks
`queue show` lists each queue's members.
*45<queue> toggles login and checks PQMSTATUS/UPQMSTATUS, so a code for
a queue the caller isn't an agent of plays ss-noservice and stores
nothing (spec snippet fixed). Pauses live in queue_member_state, AstDB
SP/QPAUSE (resynced on connect) and app_queue; QueuePause is re-sent
after every `queue reload all` and rollback, which unpause everyone.
/api/queues CRUD with members [{extensionId, penalty}], admin
pause/unpause, GET /:id/stats; /api/me/queues/:id/pause|unpause for
Home. Stats come from queue_calls (new table), filled from app_queue's
AMI events plus UserEvent SpQueueExit in the fail branch: a real call
showed that when no agent rings, the Queue part of a call vanishes into
the next leg's CDR, and Asterisk 24 sends QueueCallerAbandon on
timeouts too.
Live: QueueTracker (QueueStatus on connect and after applies, queue
events); admins see every queue, users the queues they're agents of.
Web: Queues page (list, form with every column, detail page with
waiting callers, agents and pause buttons, today's stats), queue tiles
on the dashboard, queue login switches on Home, Queue in the
DestinationPicker.
CDR fixtures for queue answered/abandoned/timeout calls. e2e m8-t1 with
real phones: 3 callers wait with 2 agents, one paused by *45550 and one
in the UI, unpause connects the longest waiting caller, one times out
to the overflow phone, stats count it all. The whole e2e chain passes.
Task: M8-T1
CallTracker groups channels by Linkedid from channel events (loaded from
CoreShowChannels on connect); call.updated/call.removed go to admins.
The dashboard lists calls now with state and timer, hang up (confirm)
and transfer to an internal number (AMI Redirect of the caller), both
audited. e2e m7-t4: 3 concurrent calls match core show channels.
Task: M7-T4
sp-sub-record now also sets CDR(SP_REC); phone numbers can record every
call (record_inbound), with the notice from Settings → Recording played
first (SP_REC_ANNOUNCE). Aggregation links the MixMonitor file as a
recordings row (size, ffprobe length). /api/recordings lists, plays
(Range; admins and the people on the call) and deletes; recordings past
recordings.retentionDays are deleted daily.
Web: Recordings page (play, download, delete), a player in Call history
and My calls, a Recording settings tab with the consent notice (08 §5),
and "Record every call" on phone numbers.
e2e m7-t3: the notice plays before a recorded inbound call and the
recording plays from Call history (the M7 acceptance). The AMI
Originate helper moved to scripts/e2e/ami.mjs.
Task: M7-T3
/api/calls (admin): filters by company-time-zone days, direction,
status, user, phone number and number search; newest first with a
keyset cursor; /:id adds the raw CDR legs; /export.csv with the same
filters (RFC 4180, spreadsheet-formula cells quoted). /api/me/calls
lists my calls with my side (in/out) and whether I missed it.
Web: admin Call history (filters, Load more, a row expands to its legs,
Export CSV), My Phone → Calls (click a number to call back: softphone
or desk phone first), and Missed calls on Home.
e2e m7-t2 with real calls found that a call going straight to voicemail
(DND) wasn't in the user's history: the mailbox owner of a VoiceMail
leg now counts as rang (and the IVR fixture's expectation follows).
Task: M7-T2
/api/blocked-numbers stores numbers normalized (any input format) and
mirrors them to AstDB SP/BLOCK/<number>, which every trunk entry checks
(Hangup 21); the AstDB resync on connect includes them. Blocked numbers
page: add with a note, unblock.
e2e m6-check runs the M6 milestone scenario with real calls into a trunk
entry: DID → office hours (closed → voicemail; *31N from an admin's phone
forces open) → menu (1 → ring group ringing two phones together, 2 → a
user), and a caller blocked in the UI is rejected while others get
through.
Task: M6-T8
/api/announcements CRUD (sound must be a prompt, destination must exist
and not be itself, delete refused while referenced); the generator
renders sp-ann-{id} (Answer, Playback, destApp) and destApp sends
announcement destinations there. The ivr golden case gains "press 3:
hours announcement, then back to the menu". Announcements are managed in
a section of Sounds & music (05 updated); the DestinationPicker gains
"Announcement".
Task: M6-T5
/api/time-conditions (company time zone by default, ranges validated,
unique override digit, never pointing at itself) with a live status
("open/closed/holiday now", same logic as the dialplan, shared), and
POST /:id/override, which writes PostgreSQL then AstDB SP/TCOVR.
/api/holidays CRUD plus import-us-federal (observed dates, computed in
code, skips days already listed). services/astdb.ts resyncs AstDB on
every AMI connect (PostgreSQL wins); services/toggles.ts applies
SpToggle TCOVR from *31N/*32N/*33N.
Generator: sp-tc-{id} per 03 §6.10 (override, holidays within the next
year, ranges rendered a minute short because GotoIfTime's end is
inclusive), sp-tc-override with a no-service fallback, admin-only
*3[123]X in sp-featurecodes, destApp time_condition. The date changing
in the company time zone triggers a forced apply so holidays roll over.
Golden cases holidays and ivr-business-hours.
Office hours page: click-and-drag half-hour grid plus an exact range
list, destinations, holiday destination, override key, status chip with
Force open / Force closed / Automatic; Holidays tab with the US federal
import. e2e m6-t4 found two bugs (pointer capture release throwing,
drawer overflow) and checks routing, AstDB, *321/*331 from an admin
phone, refusal for others, and the import reaching the dialplan.
Task: M6-T4
/api/ivrs CRUD: greeting and wrong-key sounds must be prompts, every
option destination must exist (an option may point at its own menu),
keys unique; delete drops the menu's own options before checking
references, inside a transaction so a refused delete keeps them. The
generator renders sp-ivr-{id} per 03 §6.9 (Background greeting,
WaitExten, keys sorted, i/t retry then give up, include sp-local-numbers
for direct dial), adds the number to sp-local-numbers, and destApp sends
ivr destinations there. Golden case ivr (incl. # and self-reference).
Web: Digital receptionist page with a 3x4 keypad grid (pick a key, then
its destination or "unused"), SoundPicker (pick, play, upload new right
in the form), DestinationPicker gains "Digital receptionist" and a shared
target source; useDescribeDestination labels unsaved keys.
e2e m6-t3: menu built in the UI, the greeting plays, keys 1/2 ring the
right phones, a wrong key plays pbx-invalid and repeats the greeting.
Task: M6-T3
/api/ring-groups CRUD with members as ordered memberIds (unique, must
exist), number from the ring-group range, a no-answer destination
(never itself), and delete refused while something points at it. The
generator renders sp-rg-{id} (Dial through sp-rg-member: all at once with
the total ring time, or one at a time with the per-member time, music
instead of ringing on request), adds the number to sp-local-numbers, and
destApp sends ring_group destinations there. Golden case ringgroups.
The DestinationPicker is now generic over targets picked by id and gains
"Voicemail of a user" and "Ring group" (M6-T6 completes it). Caller-name
prefixes (DIDs and ring groups) get a space before the caller's name.
Ring groups page with an ordered member list.
e2e m6-t2 found the new-group form failing on an unset no-answer
destination (fixed: defaults to hang up). It checks ring-all (both ring,
191 answers) and hunt (190 alone for 5 s, then 191) with real calls. The
generator debounce test now waits for the apply instead of a fixed 150 ms.
Task: M6-T2
/api/sounds (admin) takes the raw file, converts it with ffmpeg in one
pass to <id>.wav (8 kHz) and <id>.g722 (HD for G.722 calls) in SOUNDS_DIR,
measures it with ffprobe, and stores it as custom/<id>. Delete answers 409
with what still plays it (IVR greeting/invalid sound, announcements, the
recording notice). Sounds & music page: drop or pick a file (name
suggested from the file name), play, rename, delete. MOH classes stay in
M8. The web client and greetings share one raw-upload helper; the
api test audio helpers moved to test/audio.ts.
e2e m6-t1: drag-and-drop upload in the browser, Asterisk plays
custom/<id>, rename, delete refused while in use, then deleted.
Task: M6-T1
/api/me/voicemail lists, streams (Range, download) and marks read or
deletes through AMI VoicemailMove/VoicemailRemove, then rescans so MWI
and the list agree. /api/me/greetings takes the raw recording or file and
converts it with ffmpeg to 8 kHz PCM (media/audio.ts). My Phone gets a
layout with Home (new voicemails) and Voicemail (play marks read,
download, delete, record/upload greetings via MediaRecorder); the list
refetches on the live voicemail.new event. e2e m4-t3 checks MWI goes off
within 5 s after play and delete, and a browser-recorded greeting is what
callers hear.
Task: M4-T3
Dialplan: sp-emergency-dial (911, 9911) first in sp-from-ext;
sp-emergency has no class, channel-limit or DND check, dials every
enabled trunk in name order with the E911 caller ID, and raises
UserEvent SpEmergency. Static sp-emergency-alert (_X! so "0" matches)
for the ring-out.
api: SpEmergency -> emergency_events row, live emergency.call (admins
and notify users only), email via nodemailer (SMTP from settings) and
an Originate to every notify extension, each outcome recorded. Test
button runs the same with TEST in front, records and dials nothing.
GET/PATCH /api/settings (smtp password write-only, references
checked, read-only system keys), SMTP test, /api/system/warnings.
Fix: clearing a nullable setting stored SQL NULL (now JSON null).
UI: Settings page (Email, Emergency with the 06 §6 notice), user
Emergency tab, dashboard warnings banner, red 911 banner.
Verified: generator/emergency.test.ts; trunk-test.ts dials 911 from a
caller with no class to the fake provider (E911 caller ID in RPID,
SpEmergency on AMI, alert plays on baresip 101); scripts/e2e/m3-t4.mjs
(SMTP sink receives the test email, 101 rings, banner shows).
nodemailer pinned to 10.0.11 (newest release over two weeks old).
Task: M3-T4
Outbound routes CRUD, reorder and POST /test (mirrors the dialplan:
911 and internal numbers first, then the first matching enabled route,
class check, usable trunks, caller ID priority route > user > default).
The first trunk creates the default rules: toll-free first (because
1NXXNXXXXXX also matches 1800...), US/Canada, international disabled.
Generator: Asterisk orders patterns in one context by specificity, not
by file order, so each route gets its own sp-or-{id} context and
sp-outbound-routes includes them in position order (spec 03 §6.12
updated). sp-route-{id}: class check -> ss-noservice, caller ID
priority, per-trunk GROUP channel limit, failover on CHANUNAVAIL or
CONGESTION, all-circuits-busy-now when every trunk is full or off.
UI: Outbound rules page (↑/↓ reorder, pattern rows with help, ordered
trunk list, caller ID override, test box) and a Calling tab on users
(calling permission, outbound caller ID).
trunk-test.ts: 101 dials 3125550000 -> the fake provider receives
13125550000 with the DID in Remote-Party-ID (From user is the
sub-account), two-way audio; a caller without a class hears
ss-noservice. scripts/e2e/m3-t3.mjs covers the UI. With real voip.ms
still to do: whether per-call caller ID is honored (06 §4).
Task: M3-T3
DIDs CRUD (any input format, normalized; 409 on duplicates; a number
that is an emergency or default caller ID can't be removed). Shared
destination service: assertDestinationExists, describeDestination
("Ring Dan Doe (101)") and findReferences over every destination
column in 02 §1 plus the unmatched-number setting; deleting an
extension a destination points at is now a 409 naming the places.
Generator: destApp (user, external, hangup for now), sp-from-trunk-*
per enabled trunk, sp-did-exists (left out when empty) and sp-inbound
with caller-name prefix. Golden case voipms-basic. Phone numbers page
with a minimal DestinationPicker (user, outside number, hang up).
trunk-test.ts: the fake provider calls the DID with a 10-digit
Request-URI like voip.ms; baresip 101 answers, caller ID arrives
normalized, 500 packets each way on both legs. scripts/e2e/m3-t2.mjs
checks the UI → Asterisk path. Still to do with real voip.ms: the
mobile-to-DID call and recording which header carries the DID (06 §3).
Task: M3-T2
voip.ms-first trunk form with the portal checklist, POP server list
and Test button; passwords are write-only and encrypted. Registration
and IP-auth trunks render per 03 §5.3 (TCP/TLS add ;transport=). The
API refuses a server name that doesn't resolve and transports that are
off, because either makes Asterisk drop the trunk.
Live state: registration from PJSIPShowRegistrationsOutbound (re-read
on every Registry event: two sub-accounts on one server share
Username/Domain), reachability from PJSIPShowEndpoint and trunk
ContactStatus (static contacts aren't in PJSIPShowContacts). Trunk
state is admin-only. Dashboard gets a "Trunks up" card.
trunk-test.ts (in pnpm test:asterisk) runs a fake voip.ms (second
Asterisk): the good trunk registers, a wrong password is Rejected,
qualify reaches the provider; its recorded AMI drives the unit tests.
scripts/e2e/m3-t1.mjs does the same through the UI. The acceptance
check with a real voip.ms sub-account is still to do.
Tests: settle() now awaits an apply instead of a fixed sleep (flaked
under full-suite load).
Task: M3-T1
LiveState loads extension states and registered devices from Asterisk on
every AMI connect and follows ExtensionStatus/ContactStatus; the hub sends
a snapshot then deltas batched every 250 ms (devices for admins only),
pings every 30 s and closes unauthenticated sockets with 4401. The web UI
keeps one reconnecting socket in a zustand store, shows a Reconnecting
banner, a Dashboard (online, on a call, telephony engine) and status dots.
Fix: Asterisk 24 has no 'pjsip reload'; reload with
'module reload res_pjsip.so'. AMI errors now carry the CLI output.
Milestone check, automated: call-test.ts registers two baresip phones and
holds a 30 s call with RTP both ways (pnpm test:asterisk); the recorded
session drives the live-state tests. scripts/e2e/m2-live.mjs runs api +
Asterisk + phones + browser on a private Docker network: a new extension
is live in Asterisk in ~1.3 s, the dashboard shows 2/2 online and red dots
during the call.
Task: M2-T6
One namespace for internal numbers (range, reserved codes, parking, every
object table) with a next-free lookup. Creating an extension generates its
SIP password and voicemail PIN and a runtime-state row; every change and
password regeneration triggers a config apply. Credentials are revealed
only on request and each reveal is audited. Users get an Ext. column and a
drawer with General (number + next free), Phone and Credentials tabs.
scripts/e2e/m2-t5.mjs checks the flow in a browser and that the generated
PJSIP file follows; the files it produced load cleanly in Asterisk.
Task: M2-T5
render() is pure and gives every file the header (bodies arrive per task).
apply() runs one at a time, debounces requestApply, writes changed files
atomically (mode 0640), keeps .prev, sends each needed reload once, skips
pjsip reload and sets restartRequired when transports change, and records
generator_runs. AMI connects on startup and every reconnect forces a full
apply. Admin routes: config apply/rollback and a masked generated-files view.
Task: M2-T2
Login, forced password change, admin layout with every 05 §2 page in the
sidebar (unbuilt ones name their milestone), Users table/cards + drawer
(General tab), one-time temporary passwords, reset password. The last
active admin can't be demoted, deactivated or deleted. Audit diffs only
mask secret strings. scripts/e2e has a dependency-free CDP browser check
for this task's acceptance flow.
Task: M1-T4
argon2id passwords with a min length and a common-password list (SecLists,
MIT). Login failures are limited per IP and per email; must_change_password
blocks everything outside /api/auth/*. The first admin is created from env
on startup. pnpm dev derives DATABASE_URL from .env.
Task: M1-T3
iamdoubz