Open-source toolkit lets AI agents search award flights across 25+ loyalty programs, compare cash vs. points, and book — via MCP servers and Claude Code.
A developer released an open-source travel hacking toolkit on GitHub that integrates with Claude Code and OpenCode via MCP servers and markdown-based skills files. It connects to 25+ loyalty programs and travel APIs including Seats.aero, AwardWallet, Skiplagged, Trivago, and Airbnb. Five MCP servers work with zero API keys out of the box. The toolkit uses AI agents to answer 'burn points or pay cash?' by comparing award availability, cash prices, and loyalty balances in real time.
This toolkit is a working, forkable reference implementation of the MCP + skills-file architecture for Claude Code and OpenCode. The pattern — markdown skills files teaching the AI how to call APIs, combined with MCP servers for live tool access — is directly portable to any domain: legal research, e-commerce, SaaS tooling. The zero-API-key cold start (5 free MCP servers) is a solid DX template worth stealing.
Clone the repo and run the setup script against Claude Code this week — use it as a reference architecture to wire your own domain's APIs into an MCP server with skills files, targeting a working agent demo in under an hour.
Run: git clone https://github.com/borski/travel-hacking-toolkit.git && cd travel-hacking-toolkit && ./scripts/setup.sh
Tags
Signals by role
Also today
Tools mentioned