Ugam explained his thought process, which involved offering the Htlc output and signing the commitment transactions, broadcasting the commitment transaction containing the offered HTLC outputs to the main chain, and spending the third output using 0 0 in the witness stack, which would send the outputs to the P2WSH with to_self_delay. However, Pierre clarified that the HTLC-Timeout transaction is indeed timelocked with locktime=cltv_expiry. Ugam apologized for his misunderstanding and thanked Pierre for his help.