Travel Hacking Toolkit: AI Skills and MCP Servers for Points and Miles Search

✍️ OpenClawRadar📅 Published: April 14, 2026🔗 Source
Travel Hacking Toolkit: AI Skills and MCP Servers for Points and Miles Search
Ad

The Travel Hacking Toolkit is a collection of AI skills and MCP servers that enable AI coding agents to handle points and miles travel planning. It teaches Claude Code and OpenCode to answer the core question: "Should I burn points or pay cash?" by automating searches across multiple data sources.

Quick Setup

Clone the repository and run the setup script:

git clone https://github.com/borski/travel-hacking-toolkit.git
cd travel-hacking-toolkit
./scripts/setup.sh

The setup script walks you through choosing your tool (OpenCode, Claude Code, or both), creates API key config files, installs dependencies, and optionally installs skills system-wide. For Claude Code, launch with:

claude --strict-mcp-config --mcp-config .mcp.json

The --strict-mcp-config flag tells Claude Code to load MCP servers from the config file directly, which is more reliable than auto-discovery.

What's Included

MCP Servers (Real-time Tools)

  • Skiplagged: Flight search with hidden city fares (no API key needed)
  • Kiwi.com: Flights with virtual interlining (no API key needed)
  • Trivago: Hotel metasearch across booking sites (no API key needed)
  • Ferryhopper: Ferry routes across 33 countries, 190+ operators (no API key needed)
  • Airbnb: Search Airbnb listings with geocoding fix and property type filter (no API key needed)
  • LiteAPI: Hotel search with live rates and booking (requires LiteAPI key)

Skills (API Knowledge for Your AI)

  • duffel: Real-time flight search across airlines via Duffel API
  • seats-aero: Award flight availability across 25+ mileage programs (requires Seats.aero Pro/Partner key)
  • awardwallet: Loyalty program balances, elite status, history (requires AwardWallet Business key)
  • serpapi: Google Flights cash prices, hotels, destination discovery (requires SerpAPI key)
  • rapidapi: Secondary prices via Google Flights Live + Booking.com (requires RapidAPI key)
  • atlas-obscura: Hidden gems near any destination (no API key needed)
  • scandinavia-transit: Trains, buses, ferries in Norway/Sweden/Denmark (requires Entur + Trafiklab keys)
  • wheretocredit: Mileage earning rates by airline and booking class across 50+ frequent flyer programs (no API key needed)
Ad

How It Works

Skills are markdown files that teach your AI how to call travel APIs. They contain endpoint documentation, curl examples, useful jq filters, and workflow guidance. Both OpenCode and Claude Code support skills natively. The skills/ directory is the canonical source.

MCP servers give your AI real-time tools it can call directly. Configs are in opencode.json for OpenCode and .mcp.json for Claude Code. Five of the six MCP servers need zero API keys.

Required API Keys

For the full experience, add at minimum:

  • SEATS_AERO_API_KEY: For award flight search (Pro ~$8/mo)
  • SERPAPI_API_KEY: For cash price comparison (100 free searches/month)

Reference Data Included

The toolkit includes transfer partner ratios for Chase UR, Amex MR, Bilt, Capital One, and Citi TY. Point valuations are sourced from TPG, Upgraded Points, OMAAT, and View From The Wing. It also contains alliance membership data, sweet spot redemptions, booking windows, and hotel chain brand lookups.

📖 Read the full source: HN AI Agents

Ad

👀 See Also

TranscriptionSuite v1.1.2 adds WhisperX, NeMo, and VibeVoice models
Tools

TranscriptionSuite v1.1.2 adds WhisperX, NeMo, and VibeVoice models

TranscriptionSuite v1.1.2 now offers three transcription pipelines: WhisperX with PyAnnote diarization, NeMo models (Parakeet & Canary) with PyAnnote diarization, and VibeVoice models with built-in diarization. The update includes a model manager, parallel processing, shortcut controls, and a 24kHz recording pipeline for VibeVoice.

OpenClawRadar
Building an Agentic RAG for Obsidian with Claude and an Eval Harness to Detect Hallucinations
Tools

Building an Agentic RAG for Obsidian with Claude and an Eval Harness to Detect Hallucinations

A developer built an agentic RAG system over an Obsidian vault to let Claude answer questions from engineering books, then created an eval harness using Claude Sonnet as a judge to detect when the agent was confidently wrong. Rubric iterations improved judge-human agreement from 39% to 94%.

OpenClawRadar
Memento v1.0: Persistent Memory MCP Server for Claude Code with 17 Tools
Tools

Memento v1.0: Persistent Memory MCP Server for Claude Code with 17 Tools

Memento v1.0 is a persistent memory MCP server for Claude Code that ships with 17 tools, hybrid search, contradiction detection, and a visual memory graph. It runs locally with no cloud dependencies and supports multiple IDEs including Claude Code, Cursor, Windsurf, and OpenCode.

OpenClawRadar
Beginner-Friendly Live Build: Your First OpenClaw Agent
Tools

Beginner-Friendly Live Build: Your First OpenClaw Agent

The Clawcast hosts a live beginner-friendly build on July 8, showing how to set up an OpenClaw agent with no coding background. Join on Discord or catch the recording on YouTube.

OpenClawRadar