How manikandan-eoxys/agentichrvoicebot works
HR Voice Agent joins a LiveKit room, processes employee speech through speech recognition and a local LLM, uses HR tools and data services as needed, then returns spoken responses. The repository also includes a browser voice dashboard, call blocking, monitoring, and enterprise-data synchronization. Some domain capabilities are documented or named in the source index but their implementations were not sampled; those branches are shown without speculative internal call chains.
Main entry points: agent.py LiveKit agent · voice_controller.py Voice controller · hr_tools.py HR tool layer · sync_scheduler.py Background sync
Components 24
Voice runtime
- LiveKit agent
agent.py - Voice controller
controllers/voice_controller.py - Ollama LLM
ollama_llm.py - Deepgram STT/TTS
agent.py
HR operations
- HR tool layer
hr_tools.py - Employee tools
tools/employee_tools.py - Leave tools
tools/leave_tools.py - Leave management
services/leave_service.py - HR services
services
Data and integrations
- HR database
database/db.py - IFS Cloud client
integrations/ifs_cloud_client.py - SQL Server client
integrations/sql_server_client.py - Background sync
integrations/sync_scheduler.py
Monitoring and security
- Global monitor FSM
controllers/global_monitor_fsm.py - Monitoring handler
services/monitoring_handler.py - Call blocklist
security/call_blocklist.py - Session recorder
services/session_recorder.py
Browser dashboard
- Voice dashboard
frontend/src/components/VoiceAssistant.jsx - LiveKit client
frontend/src/services/livekitService.js - Token service
frontend/src/services/tokenService.js
Other components
- Employee
- LiveKit Cloud
- Ollama service
- Deepgram services
Connections 28
- Employee → Voice dashboard: speaks to
README.md - Voice dashboard → LiveKit client: connects through
frontend/src/components/VoiceAssistant.jsx - LiveKit client → Token service: requests token
frontend/src/services/livekitService.js - LiveKit client → LiveKit Cloud: joins room
frontend/src/services/livekitService.js - LiveKit Cloud → LiveKit agent: dispatches session
README.md - LiveKit agent → Voice controller: uses
agent.py - LiveKit agent → HR tool layer: registers tools
agent.py - LiveKit agent → Global monitor FSM: starts monitor
controllers/global_monitor_fsm.py - Voice controller → Global monitor FSM: reports state
controllers/voice_controller.py - Voice controller → Ollama LLM: builds engine
controllers/voice_controller.py - Voice controller → Background sync: starts scheduler
controllers/voice_controller.py - LiveKit agent → Deepgram STT/TTS: configures engines
README.md - Deepgram STT/TTS → Deepgram services: transcribes and speaks
README.md - Ollama LLM → Ollama service: generates response
README.md - HR tool layer → Employee tools: exposes
- Employee tools → HR database: queries
tools/employee_tools.py - HR tool layer → Leave tools: exposes
- Leave tools → Leave management: calls
tools/leave_tools.py - Leave management → HR database: reads and writes
services/leave_service.py - HR tool layer → HR services: offers HR capabilities
- Background sync → IFS Cloud client: syncs records
integrations/sync_scheduler.py - Background sync → SQL Server client: syncs records
integrations/sync_scheduler.py - Background sync → HR database: reads pending data
integrations/sync_scheduler.py - LiveKit agent → Monitoring handler: uses monitoring
agent.py - LiveKit agent → Session recorder: records sessions
agent.py - LiveKit agent → Call blocklist: checks callers
agent.py - Voice dashboard → LiveKit client: renders room events
frontend/src/components/VoiceAssistant.jsx - Global monitor FSM → Monitoring handler: pushes monitoring