How sufyansidd19/ai-travel-planner-agentic-ai works
This repository contains two travel-planning workflows: a newer LangGraph supervisor flow with specialist research, itinerary drafting, human approval and persistent state, and a simpler sequential flight/hotel workflow in the root-level app. The MCP configuration exposes Tavily, AviationStack and weather tools; the legacy flow directly uses flight and web-search helpers. README capabilities are included where documented, but the excerpts do not show every integration call or the full guardrail and approval implementation.
Main entry points: graph.py Supervisor graph · main.py Sequential travel graph · agent.py Research specialists · frontend.py Planner UI
Components 19
User interfaces
- Planner UI
Travel Agent MCP & HILP/frontend.py - Legacy UI
frontend.py - Alternate UI
UI.py
Agent orchestration
- Supervisor graph
Travel Agent MCP & HILP/graph.py - Guardrails and routing
Travel Agent MCP & HILP/agent.py - Sequential travel graph
main.py
Travel planning
- Research specialists
Travel Agent MCP & HILP/agent.py - Itinerary drafting
Travel Agent MCP & HILP/agent.py - Human approval
Travel Agent MCP & HILP/agent.py - Final plan
Travel Agent MCP & HILP/agent.py
Data integrations
- MCP tool client
Travel Agent MCP & HILP/mcp_client.py - MCP data services
- Flight search
tools/flight_tool.py - Web search
tools/tavily_tool.py - Groq LLM
State and memory
- Travel state
Travel Agent MCP & HILP/state.py - PostgreSQL checkpoints
- Legacy checkpoints
Other components
- Traveler
Connections 23
- Traveler → Planner UI: enters trip request
Travel Agent MCP & HILP/frontend.py - Planner UI → Supervisor graph: invokes graph
Travel Agent MCP & HILP/frontend.py - Supervisor graph → Guardrails and routing: routes request
Travel Agent MCP & HILP/graph.py - Supervisor graph → Research specialists: dispatches specialists
Travel Agent MCP & HILP/graph.py - Supervisor graph → Itinerary drafting: dispatches itinerary
Travel Agent MCP & HILP/graph.py - Supervisor graph → Human approval: routes for approval
Travel Agent MCP & HILP/graph.py - Supervisor graph → Final plan: routes finalization
Travel Agent MCP & HILP/graph.py - Guardrails and routing → Research specialists: selects agents
Travel Agent MCP & HILP/graph.py - Research specialists → MCP tool client: requests MCP tools
Travel Agent MCP & HILP/agent.py - MCP tool client → MCP data services: invokes tools
Travel Agent MCP & HILP/mcp_client.py - Itinerary drafting → Groq LLM: generates draft
- Final plan → Groq LLM: polishes plan
- Traveler → Human approval: provides feedback
Travel Agent MCP & HILP/agent.py - Supervisor graph → Travel state: uses state
Travel Agent MCP & HILP/graph.py - Supervisor graph → PostgreSQL checkpoints: checkpoints state
Travel Agent MCP & HILP/graph.py - Traveler → Legacy UI: enters trip request
frontend.py - Legacy UI → Sequential travel graph: invokes app
frontend.py - Traveler → Alternate UI: enters trip request
UI.py - Alternate UI → Sequential travel graph: invokes app
UI.py - Sequential travel graph → Flight search: searches flights
main.py - Sequential travel graph → Web search: searches hotels
main.py - Sequential travel graph → Groq LLM: generates itinerary
main.py - Sequential travel graph → Legacy checkpoints: checkpoints graph
main.py