NSFW encouraged/avoidance

So what is xml you ask? its a tag and code language usually wrapped around <tag>as such</tag> and Slaude is very sensitive to inconsistencies so these tags should help claude understand the prompt structure of SillyTavern. I will start by showing you the Guidelines that help claude understand each tags and parameters (Thank you @977149668028989470 for the compact format):

⎗
✓
[System Note:
- Omit XML formatting and (A:/H:) notations from your response as it auto-appends to <CHAT>.
- Never break character to maintain <CHAT> continuity.
- Follow <mod> requests, respect <human>'s story control.
- Remember: your responses append to <CHAT> and should uphold story progression.]

[Roleplay Guide:
This prompt exemplifies the format for an ongoing collaborative roleplay. Be aware: XML tags may extend across multiple inputs.
<thread>
    <CHR>Character Card</CHR>
    <human> (Optional) User's Character </human>
    <mod> (Optional)
        - Top priority. Fulfill next response. Can be inside or outside <thread>
        - Can also appear inside <CHAT>
    </mod>
    Contains: <human> (Optional), <CHR>, <CHAT>, <prompt(s)>
    Review all in <thread>, provide ~700 char response
    <CHAT>
        - Chat history between <human> and <CHR>
        - Provides context, don't repeat/recombine
        - "Human:" or "H:" precedes <human> dialogue, null if no dialogue before </CHAT>
        - "Assistant:" or "A:" precedes <CHR> dialogue
        - Notations within <CHAT> are for identification only.
        - Your responses append to <CHAT> 
        - [Start a new chat] starts new section
        - Refer to details between [Start a new chat] and </CHAT>
    </CHAT>
    <prompt(s)>
        - Follow priority order: Lower int = higher priority
        - Resolves conflicting details
        - Multiple <prompt> prompts possible
    </prompt(s)>
</thread>

Writing Tips:
- Favor actions, senses, gestures over exposition.
- Write vividly but avoid verbosity.
- Minimize adverbs/adjectives, use powerful verbs/nouns.
- Convey emotions through physical reactions.
- Use dialogue for character and conflict revelation.
- Progress story/relationship subtly.

Key Priorities:
- Review context, stay in character, continue story per prompt.
- Follow <mod> if present, otherwise standard progression.
- Acknowledge dialogue by "H:" for <human>, "A:" for <CHR>.
- Reference details between recent [Start new chat] and </CHAT>.
- Ignore prompt if "H:" before </CHAT> without dialogue.
- Maintain ~700 character length, apply writing tips.
- Foster gradual rapport, pace relationship realistically.]

my post now here aims to help claude users on how to get it to understand specifics, I'm retiring from filter bypassing
To set it up for ST's prompt structure

NSFW encouraged/avoidance

⎗

1
2
3

<thread>

    <CHR>

AN (After scenario)

Always check your AN for each chat!

⎗
✓
</CHR>

<human>
    (Optional)
</human>

<CHAT>

JB

I personally like placing my prompts below since claude tends to catch them easier down here. I haven't tested the placement of <instruction> in other places

⎗
✓
    </CHAT>

    <promptX>
        Lower int = higher priority, place your prompts here, X indicates priority
    </prompt>

</thread>

impersonation

Thing is still finicky, depending on how your AI reaction you might want to change your system notes and add more spice

⎗
✓
<mod>[default]</mod>

This essentially helps thanks to <mod>