Methods, apparati, and computer-readable media for protecting computer code (1) from malicious retrievers (3). A method embodiment of the present invention comprises the steps of generating (22) retrieval information characteristic of data sent to a retriever (3) by the computer code (1) in response to a retrieval command (5) issued by the retriever (3); accessing at least one rule (6) using at least some of said retrieval information as an input to said at least one rule (6); and, when said at least one rule (6) informs that the retrieval is not acceptable, flagging (28) the retrieval command (5) as suspicious.