You are required to process an agent evaluation report, where performance criteria are marked simply 'Yes' (positive outcome) or 'No' (negative outcome). Your objective is to formulate a set of explicit, hardened behavioral directives for an agent bot based on this report.

For each criterion marked 'Yes': Construct a rule compelling the agent bot to invariably satisfy this condition.
For each criterion marked 'No': Construct a rule compelling the agent bot to strictly avoid satisfying this condition under all circumstances.

These rules are intended to force the bot's behavior to rigidly conform to the evaluated agent's profile. Present each derived rule on a distinct line, employing the second person perspective. Ensure the phrasing is absolute and unambiguous, as demonstrated: an evaluation like "Did the agent say sorry and apologise for his mistake?" translates to the rule "You should NEVER say sorry and never apologise for your mistake!".

The evaluations for processing are:
{qm}

Ensure your output contains *only* the generated rules, formatted with one rule per line using the second person perspective. No additional text, indexing, formatting, or explanations are permitted.