=== Role-playing instruction (fixed template), without the output-format sentence
    paper: Tables/prompts_agent.tex:11-47
    platform: CHARACTER_PROMPT.prompt + othersSection + motivationSection + requirements
    tokens: paper 33, platform 22
    delete: paper[17:19] '(if available)'  ->  platform[17:17] ''
    delete: paper[23:32] '(if available) ===Relevant Background Information=== {retrieved_knowledge} (if retrieval augmented)'  ->  platform[21:21] ''

=== Output format, CoSER models
    paper: Tables/prompts_agent.tex:52
    platform: CHARACTER_PROMPT.outputFormat.thought
    tokens: paper 26, platform 26
    IDENTICAL (token sequence)

=== Output format, other models
    paper: Tables/prompts_agent.tex:58
    platform: CHARACTER_PROMPT.outputFormat.thoughtWithExample
    tokens: paper 45, platform 45
    IDENTICAL (token sequence)

--- Text the platform sends to a character that is in no paper table (AGENT.concise): '\n\nSpeak concisely as humans, instead of being verbose. Limit your response to 60 words.\n\n'
--- Text the platform sends to name-prefix families that is in no paper table (AGENT.namePrefix): 'Start your response with "{name}: ". Avoid speaking as other characters.\n\n'
--- Separators between other-character profiles (CHARACTER_PROMPT.decorators), not in the paper table: ['******************************\n\n', '********************\n\n', '\n\n', '\n', '']

=== Environment model
    paper: Tables/prompts_agent_2.tex:11-36
    platform: ENVIRONMENT_PROMPT.prompt, with the random role fixed to the paper's "an environment simulator"
    tokens: paper 153, platform 153
    IDENTICAL (token sequence)
    note: platform draws the role from ['an environment model', 'a world model', 'a world simulator', 'an environment simulator'] (paper table prints only "an environment simulator")

=== Next speaker, the prompt the platform USES
    paper: Tables/prompts_agent_2.tex:42-46
    platform: NSP_PROMPT.prompt (used by setup.ts:133)
    tokens: paper 76, platform 103
    insert: paper[76:76] ''  ->  platform[76:103] '===Output Format=== You must structure your response as follows: 1. Reasoning: Explain your thought process. 2. Next Speaker: Select exactly one character name from the list: {all_characters}.'

=== Next speaker, the recorded paper-era prompt (NOT used)
    paper: Tables/prompts_agent_2.tex:42-46
    platform: PAPER_ERA_NSP_PROMPT.prompt
    tokens: paper 76, platform 76
    IDENTICAL (token sequence)

=== Judge template
    paper: Tables/prompts_eval.tex:9-82
    platform: JUDGE.template
    tokens: paper 262, platform 233
    delete: paper[0:2] '\\textbf«Template» &'  ->  platform[0:0] ''
    replace: paper[55:56] '«major_characters».'  ->  platform[53:54] '{major_characters}.'
    replace: paper[61:62] '«book»'  ->  platform[59:60] '{book}'
    replace: paper[79:80] '{dimension_intro}.'  ->  platform[77:78] '{dimension_brief}.'
    replace: paper[123:152] '(if k>0) Please note that the first {k} messages in the simulated conversation are the same as the reference. Focus your evaluation only on the content after these messages.'  ->  platform[121:122] '{additional_instructions}'
    replace: paper[184:185] '{dimension_rubrics}'  ->  platform[154:155] '{dimension_criteria}'
    replace: paper[258:259] '},},'  ->  platform[228:230] '}, },'

=== Anthropomorphism: intro
    paper: Tables/prompts_eval_2.tex:13
    platform: JUDGE.details[Anthropomorphism].dimension_brief
    tokens: paper 7, platform 7
    IDENTICAL (token sequence)

=== Anthropomorphism: rubrics
    paper: Tables/prompts_eval_2.tex:19-48
    platform: JUDGE.details[Anthropomorphism].dimension_criteria
    tokens: paper 119, platform 119
    IDENTICAL (token sequence)

=== Character Fidelity: intro
    paper: Tables/prompts_eval_2.tex:55
    platform: JUDGE.details[Character Fidelity].dimension_brief
    tokens: paper 11, platform 11
    IDENTICAL (token sequence)

=== Character Fidelity: rubrics
    paper: Tables/prompts_eval_2.tex:61-89
    platform: JUDGE.details[Character Fidelity].dimension_criteria
    tokens: paper 154, platform 154
    replace: paper[9:10] '«major_characters»)'  ->  platform[9:10] '{major_characters})'

=== Storyline Quality: intro
    paper: Tables/prompts_eval_2.tex:96
    platform: JUDGE.details[Storyline Quality].dimension_brief
    tokens: paper 10, platform 10
    IDENTICAL (token sequence)

=== Storyline Quality: rubrics
    paper: Tables/prompts_eval_2.tex:102-115
    platform: JUDGE.details[Storyline Quality].dimension_criteria
    tokens: paper 59, platform 59
    IDENTICAL (token sequence)

=== Storyline Consistency: intro
    paper: Tables/prompts_eval_2.tex:122
    platform: JUDGE.details[Storyline Consistency].dimension_brief
    tokens: paper 18, platform 18
    IDENTICAL (token sequence)

=== Storyline Consistency: rubrics
    paper: Tables/prompts_eval_2.tex:128-132
    platform: JUDGE.details[Storyline Consistency].dimension_criteria
    tokens: paper 24, platform 24
    IDENTICAL (token sequence)