In LLM solicitations the agency requests at minimum the acceptable use policy, model, system or data cards, end user resources and a mechanism for end user feedback, obtaining the information through resellers or integrators where the developer is not the vendor and avoiding demands for sensitive technical data such as model weights. Where its planned use warrants, it requests enhanced information focused on the principles: pre-training and post-training actions affecting factuality and grounding, system-level prompts, outputs restricted by content moderation and safety filters, red teaming for bias, training outside the United States, and modifications made for foreign regulation; bias evaluations and benchmark scores; enterprise controls such as customisable system instructions, evaluation tools and source citation features; and third-party modifications by non-developer vendors. The agency identifies these requirements as material to eligibility and payment so that the contract can be terminated for default when a vendor refuses corrective action.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.