What is the primary goal of using stop sequences and message prefilling in this exercise?	To receive multiple commands in a single response without any additional comments or explanation.
What specific sequence was used initially to signal the start and end of a code block?	Three backticks (```).
Why did the LLM sometimes insert a language identifier (e.g., `bash`) inside the code block?	Because the initial prefill indicated a markdown code block, and the LLM chose a syntax highlighting language.
How can you prevent the LLM from adding an unwanted language identifier within a code block?	By explicitly including the desired language identifier in the pre-filled message.
What was the second problem encountered after using a bash code block?	The LLM might insert bash-formatted explanatory comments (e.g., "This command does X, Y, Z").
How can you ensure the LLM provides only the commands and no extra commentary?	By using the message prefill to provide strong, explicit instructions (e.g., "here are all three commands in a single block without any comments.").
What is a key principle regarding the use of message prefilling?	It is not limited to using specific characters; it can be used to dramatically guide the LLM's response.