llms-full.txt is a fuller documentation bundle used in some llms.txt ecosystems. It typically groups more complete site guidance, references, or installation details so AI agents and other automated tools can consume curated instructions without crawling the whole site.
In security terms, that convenience creates trust risk. If an automation workflow treats names, package references, or setup steps in llms-full.txt as authoritative, an attacker may try naming-collision or dependency-compromise abuse elsewhere and wait for the tool to follow those instructions. This makes the file part of the attack surface, not just documentation. Defenders should review it like code, keep it under change control, and prevent external names from flowing directly into install or execution steps. Used carefully, it can improve agent navigation; used loosely, it can help attackers steer automation.