Atlas, a global travel-tech company, has announced the open-source release of its Atlas Flight Booking Skill, designed to enhance AI agents with real-time flight search and controlled booking capabilities. Available on GitHub and PyPI, this tool allows developers to integrate flight-booking functions without building from scratch, addressing a common gap in AI travel agents' capabilities.
The Skill is built around four critical checkpoints requiring human approval: authorisation, price changes, seat fallback, and final payment summary. This approach ensures that AI agents do not operate autonomously in transactions, maintaining human oversight in crucial decision-making moments. Mary Li, CEO and Founder of Atlas, stated, “We open-sourced our Flight Booking Skill for AI agents — not because we wanted agents to book flights fully autonomously, but because we didn't.”
The release follows Atlas's expansion of low-cost carrier connectivity to independent travel sellers and developers. The Skill supports core booking operations, including search, pricing, availability, and booking creation. Future updates may include refunds, cancellations, and other post-booking operations.
Designed for developers, AI product teams, and agent builders, the Skill integrates with existing Atlas ATRIP accounts without requiring separate commercial agreements. Atlas continues to redefine low-cost carrier retailing by unifying content from over 140 carriers and leveraging analytics from over 1 billion daily searches
This story was selected and published by a human editor, with content adapted from original press material using AI tools and checked by the editor before publication. Spot an error? Report it here.
