de5a2abdeb
M5: hardware integration layer (write-only, live test deferred)
...
- GsmGatewaySmsProvider: GoIP-style HTTP SMS API, per-tenant SIM line from
phone_numbers.gateway_port; model-specific bits isolated for easy adaptation
(§17.3); failure never breaks the calling flow (logged to sms_log)
- Asterisk config generator now emits [gw-line-N] SIP trunk endpoints per
assigned SIM, routed straight into the tenant's inbound context (1 SIM =
1 salon §5.1); deterministic per-SIM SIP secrets
- docs/HARDWARE_TESTING.md: step-by-step live runbook for the joint session
(gateway registration, inbound path, SMS via gateway, operator forwarding
codes per m:tel/BH Telecom/HT Eronet, e2e acceptance, rollback)
- 5 tests against a mocked gateway API
(forwarding-code generator and SIM/port admin UI shipped in M3)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com >
2026-07-11 11:09:28 +02:00
714aa1782b
M4: voice pipeline (software-only telephony)
...
- AudioSocket server (asyncio TCP, Asterisk wire protocol) bridging calls
into the same M2 agent used by chat
- Call session engine: greeting → energy-VAD utterance collection → STT →
agent turn → TTS playback with barge-in (120ms of caller speech stops
playback); inactivity + max-duration guards; unintelligible audio reaches
the agent as '[nerazumljivo]' so the two-attempt rule stays in the prompt
- STTProvider (faster-whisper, lang hint 'sr', GPU/CPU via env) and
TTSProvider (Azure Neural raw-8k PCM / ElevenLabs Flash) + test fakes
- Recording (mixed caller+agent WAV), transcripts, per-turn latency trace,
metering via record_agent_call (§12)
- Internal dialplan API: /internal/calls/register (ring targets computed from
working hours + ring settings §5.2), /answered (human outcome, no minutes),
/hangup (abandoned → missed-call SMS §5.5); shared-token auth
- Asterisk config generator (pjsip.conf + extensions.conf from DB): worker
endpoints, per-tenant test-caller endpoint, register→Dial→AudioSocket
dialplan with h-extension reporting — validated against a real Asterisk 20
container (modules, dialplan, endpoints all load)
- docker-compose 'voice' profile (voice server + Asterisk), Dockerfile.voice,
docs/VOICE_TESTING.md softphone runbook
- 16 new tests incl. full fake-call e2e and a real-TCP AudioSocket wire test
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com >
2026-07-11 11:05:56 +02:00
646917c322
M3: chat widget, owner dashboard, super-admin panel
...
- Embeddable vanilla-JS chat widget (one script tag + tenant public key):
WebSocket, resumable sessions (24h localStorage token), honeypot + per-IP
and per-message rate limits, Bosnian UI (§7)
- Session auth (bcrypt + signed cookies), owner accounts + super-admins,
admin impersonation ('otvori kao salon', §11)
- Owner dashboard (§10, Bosnian): requests with same actions as email,
call/chat history with transcripts, usage bar; settings for profile,
working hours (manual + LLM paste-to-parse), services CRUD + 'Zalijepi
cjenovnik' LLM import with review step, scheduling (Google OAuth free/busy
read-only + calendar mappings + buffers, partner status), telephony
(forwarding MMI codes per operator, ring group, worker SIP creds + QR),
agent voice/price-mode/notes + widget snippet, notifications/SMS templates
- Super-admin panel (§11): tenant CRUD incl. plan/minutes/paid-until,
partner API config (encrypted secrets), number/SIM registry + assignment,
connectivity test (live availability + dry-run push), versioned prompt
template editor with publish/rollback, playground, usage overview, health
- Fixes: WAL + busy_timeout for sqlite tests, no awaits in WS disconnect
path (deadlock), template publish autoflush bug
- 76 tests green (incl. widget WS e2e booking flow); dashboard, widget and
admin verified live in a browser against Postgres
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com >
2026-07-11 10:51:57 +02:00
0d0838e6e2
M2: agent logic in text mode
...
- LLMProvider interface: AnthropicLLM + ScriptedLLM test fake
- System prompt composition from super-admin template + structured tenant
data only (§6.3); auto-generated greetings with recording disclosure,
out-of-hours variant, chat variant
- Agent tools (§6.5): get_salon_info, check_availability (≤8 slots to model,
agent offers ≤3), submit_booking_request (caller-ID fallback, hallucinated
service-id guard), take_message
- Conversation loop with tool dispatch, round-limit guard, transcript capture,
outcome classification (request_created/message_taken/info_only/abandoned)
- Terminal playground (python -m gogo.agent.cli) + demo seed (python -m gogo.seed)
- Deterministic loop tests (ScriptedLLM) + Appendix A golden-scenario
acceptance tests against a real LLM (skipped without an API key)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com >
2026-07-11 09:53:08 +02:00
e855650f09
M1: backend core + proposal engine
...
- FastAPI skeleton, SQLAlchemy models (§13), Alembic initial migration
- SchedulingProvider interface with google_calendar (free/busy read-only),
partner_api (Appendix B client) and mock implementations
- Proposal engine: create → provider-routed delivery → owner actions
(resolve/confirm+SMS/reject) → expiry + reminders (§9)
- Signed single-use action links, .ics METHOD:REQUEST attachment
- Partner outcome webhook with HMAC verification + polling fallback
- SmsProvider (console) with Bosnian templates (§5.5), EmailProvider (console/SMTP)
- Fake partner API server in tests/ — Appendix B reference implementation
- 43 tests: slot math, proposal lifecycle, action links, partner contract
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com >
2026-07-11 09:45:06 +02:00
