Tejasurisetti

Travelpilot — AI skill for Claude Code

AI community

AI-powered trip planning agent for Indian state capitals — builds day-by-day itineraries, detects scheduling conflicts, and rebuilds plans in real time when bookings or activities change.

How to install Travelpilot

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Tejasurisetti/travelpilot and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Travelpilot does

AI-powered trip planning agent for Indian state capitals — builds day-by-day itineraries, detects scheduling conflicts, and rebuilds plans in real time when bookings or activities change. Built with Next.js, TypeScript, and Claude tool-calling.

Alternatives in AI

Tokentab — A CLI that reads Claude Code, Codex, and Gemini CLI session logs and works out how much they cost, by model, p 226 ★
Pingcrm — Personal Networking CRM — AI-powered, open-source, self-hostable 115 ★
Code Conductor — Ship 10x faster by running multiple Claude Code sub agents in parallel 110 ★

README

TravelPilot

TravelPilot is an AI-powered trip planning and disruption-management agent for a hackathon demo. It builds a day-by-day itinerary, reasons about time and location, tracks estimated spend, and rebuilds the affected part of a trip when plans change.

Problem Framing

Most travel products are good at individual searches and bookings, but they treat an itinerary as a collection of disconnected reservations. When a venue closes, a flight is delayed, weather changes, or a booking disappears, the traveler still has to manually work out which activities conflict, what is nearby, how much time remains, and whether the revised plan still fits the budget.

TravelPilot treats the itinerary as a connected plan. A change to one item can trigger deterministic conflict checks, backup selection, reordering, and budget recalculation. Claude provides the natural-language interface and chooses the right operation; TypeScript functions make the operational decisions quickly and repeatably.

Architecture

flowchart TD
  UI[Dashboard + ChatPanel] --> PLAN[POST /api/plan]
  UI --> CHAT[POST /api/chat]
  UI --> DISRUPT[POST /api/disrupt]
  UI --> REPLAN[POST /api/replan]
  PLAN --> PLANNER[lib/planner.ts]
  REPLAN --> PLANNER
  CHAT --> CLAUDE[lib/claude.ts]
  CLAUDE --> TOOLS[Tool definitions]
  TOOLS --> SCHEDULER[lib/scheduler.ts]
  DISRUPT --> SCHEDULER
  PLANNER --> SCHEDULER
  PLANNER --> MOCK[lib/mockData.ts]
  SCHEDULER --> TYPES[lib/types.ts]
  CLAUDE --> TYPES

LLM orchestrator

src/lib/claude.ts creates the Anthropic client from ANTHROPIC_API_KEY.
Claude receives the trip JSON and the traveler question.
Tool calls are dispatched to deterministic scheduler functions.
runAgentTurn returns the final answer and an auditable list of tool names and arguments. The UI displays that tool trace as operational reasoning, not hidden chain-of-thought.