How praneeth-ram/tickora-ai works
Tickora AI is a FastAPI attendance-and-HR chatbot that turns employee questions into read-only, permission-scoped database answers. The main workflow authenticates the employee, enters a LangGraph chat flow, interprets and corrects a query plan, validates access, executes an approved query, and formats a grounded response. The README also describes attendance, leave, work-from-home, timesheet, holiday, and schedule capabilities. Several endpoint and execution details are unsampled, so uncertain links are omitted or marked dashed.
Main entry points: main.py FastAPI endpoints · llm_chat_service.py Plan correction and answers · nodes.py Chat flow nodes · dynamic_query_service.py Dynamic query execution
Components 22
API and identity
- FastAPI endpoints
backend/main.py - Employee authentication
backend/dependencies/auth.py - JWT and password service
backend/services/auth_service.py - Role access policy
backend/security/access_policy.py
Chat orchestration
- Chat graph
backend/agents/graph.py - Chat flow nodes
backend/agents/nodes.py
Query planning
- Plan inference model
backend/services/model_inference.py - Plan correction and answers
backend/services/llm_chat_service.py - Date resolution
backend/services/date_resolver.py - Approved operation registry
backend/services/operation_registry.py - Answer construction
backend/services/answer_builder.py
Data access
- Schema catalog
backend/services/schema_catalog.py - Dynamic query execution
backend/services/dynamic_query_service.py - Database engines
backend/db.py
HR data domains
- Attendance records
backend/repositories/attendance_repository.py - Leave requests
backend/repositories/leave_repository.py - Work-from-home requests
backend/repositories/wfh_repository.py - Timesheets
backend/repositories/timesheet_repository.py - Holidays and schedules
backend/repositories/calendar_repository.py
Other components
- Employee
- Tickora application
- PostgreSQL
Connections 26
- Employee → Tickora application: asks question
- Tickora application → FastAPI endpoints: sends chat request
- FastAPI endpoints → Employee authentication: authenticates request
backend/main.py - Employee authentication → JWT and password service: validates token
backend/dependencies/auth.py - FastAPI endpoints → Role access policy: resolves actor
backend/main.py - FastAPI endpoints → Chat graph: invokes chat flow
backend/main.py - Chat graph → Chat flow nodes: dispatches nodes
backend/agents/graph.py - Chat flow nodes → Plan correction and answers: corrects and answers
backend/agents/nodes.py - Chat flow nodes → Date resolution: resolves dates
backend/agents/nodes.py - Chat flow nodes → Role access policy: reads subject
backend/agents/nodes.py - Plan correction and answers → Plan inference model: requests plan
backend/services/llm_chat_service.py - Plan correction and answers → Approved operation registry: loads operations
backend/services/llm_chat_service.py - Plan correction and answers → Dynamic query execution: executes plan
backend/services/llm_chat_service.py - Plan correction and answers → Answer construction: builds response
backend/services/llm_chat_service.py - Approved operation registry → Schema catalog: discovers schema
backend/services/operation_registry.py - Dynamic query execution → Approved operation registry: checks operations
backend/services/dynamic_query_service.py - Dynamic query execution → Role access policy: enforces scope
backend/services/dynamic_query_service.py - Dynamic query execution → Database engines: opens connection
backend/services/dynamic_query_service.py - Schema catalog → Database engines: inspects database
backend/services/schema_catalog.py - Database engines → PostgreSQL: connects
backend/db.py - FastAPI endpoints → Attendance records: uses repository
backend/main.py - FastAPI endpoints → Leave requests: uses repository
backend/main.py - FastAPI endpoints → Work-from-home requests: uses repository
backend/main.py - FastAPI endpoints → Timesheets: uses repository
backend/main.py - FastAPI endpoints → Holidays and schedules: uses repository
backend/main.py - Answer construction → Tickora application: returns answer