A method includes obtaining input to modify a policy of a set of context-aware document policies. A policy of the set is applicable to a requested action on a document so as to indicate allowability of the requested action based at least on satisfaction of a condition of the policy that relates to a content of the document. When a plurality of policies of the set are applicable to the requested action on the document, allowability of the requested action is determined by the allowability that is indicated by application of the applicable policy with a highest priority. The modified policy is compared with another policy of the set. If the comparison indicates the modified policy and the other policy are applicable to a single requested action on a single document, the set of policies is automatically ensured to remain self consistent.