When documenting the algorithms used in the AI system the organization should consider: whether the chosen algorithms suit the business objectives and the AI task; where the algorithms and any modifications come from (research papers, journals, presentations, source code repositories); how valid the algorithm is, for example whether recognized experts have tested and accepted it; how the algorithms have been used before and how well they worked, including whether systems based on them were deployed or withdrawn and whether they caused harm; the algorithms' data requirements; and their tendency to reach unwanted conclusions such as spurious correlations, overfitting or propagated bias. Deployers and users may not have all this information, but the assessment still has to address the AI system as a whole.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.