2026-01-02 17:11:58 +02:00
# utils/evil_mode.py
"""
Evil Mode module for Miku .
Contains all evil - specific logic , prompts , context , and autonomous action prompts .
This module is the central hub for Evil Miku ' s alternate behavior.
"""
import os
import random
import json
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
import time
import asyncio
import discord
2026-01-02 17:11:58 +02:00
import globals
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
from utils . logger import get_logger
logger = get_logger ( ' persona ' )
2026-01-02 17:11:58 +02:00
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# Evil mood rotation interval (2 hours in seconds)
EVIL_MOOD_ROTATION_INTERVAL = 7200
# Background task handle for the rotation timer
_evil_mood_rotation_task = None
2026-01-02 17:11:58 +02:00
# ============================================================================
# EVIL MODE PERSISTENCE
# ============================================================================
EVIL_MODE_STATE_FILE = " memory/evil_mode_state.json "
2026-01-02 19:45:56 +02:00
def save_evil_mode_state ( saved_role_color = None ) :
""" Save evil mode state to JSON file
Args :
saved_role_color : Optional hex color string to save ( e . g . , " #86cecb " )
"""
2026-01-02 17:11:58 +02:00
try :
2026-01-02 19:45:56 +02:00
# Load existing state to preserve saved_role_color if not provided
existing_saved_color = None
if saved_role_color is None and os . path . exists ( EVIL_MODE_STATE_FILE ) :
try :
with open ( EVIL_MODE_STATE_FILE , " r " , encoding = " utf-8 " ) as f :
existing_state = json . load ( f )
existing_saved_color = existing_state . get ( " saved_role_color " )
except :
pass
2026-01-02 17:11:58 +02:00
state = {
" evil_mode_enabled " : globals . EVIL_MODE ,
2026-01-02 19:45:56 +02:00
" evil_mood " : globals . EVIL_DM_MOOD ,
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
" saved_role_color " : saved_role_color if saved_role_color is not None else existing_saved_color ,
" last_rotation_time " : getattr ( globals , ' EVIL_LAST_ROTATION_TIME ' , time . time ( ) )
2026-01-02 17:11:58 +02:00
}
with open ( EVIL_MODE_STATE_FILE , " w " , encoding = " utf-8 " ) as f :
json . dump ( state , f , indent = 2 )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( f " Saved evil mode state: { state } " )
2026-01-02 17:11:58 +02:00
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to save evil mode state: { e } " )
2026-01-02 17:11:58 +02:00
def load_evil_mode_state ( ) :
""" Load evil mode state from JSON file """
try :
if not os . path . exists ( EVIL_MODE_STATE_FILE ) :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . info ( f " No evil mode state file found, using defaults " )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
return False , " evil_neutral " , None , time . time ( )
2026-01-02 17:11:58 +02:00
with open ( EVIL_MODE_STATE_FILE , " r " , encoding = " utf-8 " ) as f :
state = json . load ( f )
evil_mode = state . get ( " evil_mode_enabled " , False )
evil_mood = state . get ( " evil_mood " , " evil_neutral " )
2026-01-02 19:45:56 +02:00
saved_role_color = state . get ( " saved_role_color " )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
last_rotation_time = state . get ( " last_rotation_time " , time . time ( ) )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( f " Loaded evil mode state: evil_mode= { evil_mode } , mood= { evil_mood } , saved_color= { saved_role_color } " )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
return evil_mode , evil_mood , saved_role_color , last_rotation_time
2026-01-02 17:11:58 +02:00
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to load evil mode state: { e } " )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
return False , " evil_neutral " , None , time . time ( )
2026-01-02 17:11:58 +02:00
def restore_evil_mode_on_startup ( ) :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
""" Restore evil mode state on bot startup (without changing username/pfp).
Returns True if evil mode was restored , False otherwise .
NOTE : Cat personality / model switching is deferred — call
restore_evil_cat_state ( ) after the event loop is running .
"""
evil_mode , evil_mood , saved_role_color , last_rotation_time = load_evil_mode_state ( )
2026-01-02 17:11:58 +02:00
if evil_mode :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( " Restoring evil mode from previous session... " )
2026-01-02 17:11:58 +02:00
globals . EVIL_MODE = True
globals . EVIL_DM_MOOD = evil_mood
globals . EVIL_DM_MOOD_DESCRIPTION = load_evil_mood_description ( evil_mood )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
globals . EVIL_LAST_ROTATION_TIME = last_rotation_time
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . info ( f " Evil mode restored: { evil_mood } " )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# Start the rotation timer (will handle catch-up if time has passed)
start_evil_mood_rotation ( )
2026-01-02 17:11:58 +02:00
else :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
globals . EVIL_LAST_ROTATION_TIME = time . time ( )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . info ( " Normal mode active " )
2026-01-02 17:11:58 +02:00
return evil_mode
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
async def restore_evil_cat_state ( ) :
""" Switch Cat to the correct personality plugin + LLM model based on evil mode state.
Must be called after the event loop is running ( e . g . , in on_ready ) .
2026-03-01 00:57:13 +02:00
Waits for Cat to become reachable , then retries plugin switching with
verification to handle the common race condition where bot starts before Cat .
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
"""
try :
from utils . cat_client import cat_adapter
if not globals . USE_CHESHIRE_CAT :
return
2026-03-01 00:57:13 +02:00
# Wait for Cat to actually be reachable before attempting any API calls
if not await cat_adapter . wait_for_ready ( max_wait = 120 , interval = 5 ) :
logger . error ( " Cat never became ready — cannot restore personality state " )
return
# Small extra delay to let Cat fully initialize plugins after health endpoint is up
await asyncio . sleep ( 3 )
max_retries = 3
retry_delay = 5
for attempt in range ( 1 , max_retries + 1 ) :
try :
if globals . EVIL_MODE :
if attempt == 1 :
logger . info ( " Restoring Cat evil personality state on startup... " )
else :
logger . info ( f " Retry { attempt } / { max_retries } : restoring Cat evil personality... " )
await cat_adapter . switch_to_evil_personality ( )
else :
active = await cat_adapter . get_active_plugins ( )
if " evil_miku_personality " in active :
logger . info ( " Evil plugin still active after normal restore — switching to normal... " )
await cat_adapter . switch_to_normal_personality ( )
else :
# Normal mode, normal plugins — nothing to do
return
# Verify the switch actually worked
await asyncio . sleep ( 2 )
active = await cat_adapter . get_active_plugins ( )
if globals . EVIL_MODE :
if " evil_miku_personality " in active and " miku_personality " not in active :
logger . info ( " ✅ Cat evil personality verified active " )
return
else :
logger . warning ( f " Cat plugin verification failed (attempt { attempt } ): "
f " evil_active= { ' evil_miku_personality ' in active } , "
f " normal_active= { ' miku_personality ' in active } " )
else :
if " miku_personality " in active and " evil_miku_personality " not in active :
logger . info ( " ✅ Cat normal personality verified active " )
return
else :
logger . warning ( f " Cat plugin verification failed (attempt { attempt } ) " )
except Exception as e :
logger . error ( f " Cat personality restore attempt { attempt } error: { e } " )
if attempt < max_retries :
await asyncio . sleep ( retry_delay )
logger . error ( f " Failed to restore Cat personality after { max_retries } attempts " )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
except Exception as e :
logger . error ( f " Failed to restore Cat personality state on startup: { e } " )
2026-01-02 17:11:58 +02:00
# ============================================================================
# EVIL MODE CONTEXT AND PROMPTS
# ============================================================================
def get_evil_miku_lore ( ) - > str :
""" Load the evil_miku_lore.txt file """
try :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
with open ( " persona/evil/evil_miku_lore.txt " , " r " , encoding = " utf-8 " ) as f :
2026-01-02 17:11:58 +02:00
return f . read ( )
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to load evil_miku_lore.txt: { e } " )
2026-01-02 17:11:58 +02:00
return " ## EVIL MIKU LORE \n [File could not be loaded] "
def get_evil_miku_prompt ( ) - > str :
""" Load the evil_miku_prompt.txt file """
try :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
with open ( " persona/evil/evil_miku_prompt.txt " , " r " , encoding = " utf-8 " ) as f :
2026-01-02 17:11:58 +02:00
return f . read ( )
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to load evil_miku_prompt.txt: { e } " )
2026-01-02 17:11:58 +02:00
return " ## EVIL MIKU PROMPT \n [File could not be loaded] "
def get_evil_miku_lyrics ( ) - > str :
""" Load the evil_miku_lyrics.txt file """
try :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
with open ( " persona/evil/evil_miku_lyrics.txt " , " r " , encoding = " utf-8 " ) as f :
2026-01-02 17:11:58 +02:00
return f . read ( )
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to load evil_miku_lyrics.txt: { e } " )
2026-01-02 17:11:58 +02:00
return " ## EVIL MIKU LYRICS \n [File could not be loaded] "
def get_evil_complete_context ( ) - > str :
""" Returns all essential Evil Miku context """
return f """ ## EVIL MIKU LORE (Complete)
{ get_evil_miku_lore ( ) }
## EVIL MIKU PERSONALITY & GUIDELINES (Complete)
{ get_evil_miku_prompt ( ) }
## EVIL MIKU SONG LYRICS (Complete)
{ get_evil_miku_lyrics ( ) } """
def get_evil_context_for_response_type ( response_type : str ) - > str :
""" Returns appropriate evil context based on the type of response being generated """
core_context = f """ ## EVIL MIKU LORE (Complete)
{ get_evil_miku_lore ( ) }
## EVIL MIKU PERSONALITY & GUIDELINES (Complete)
{ get_evil_miku_prompt ( ) } """
if response_type in [ " autonomous_general " , " autonomous_tweet " , " dm_response " ,
" server_response " , " conversation_join " ] :
return f """ { core_context }
## EVIL MIKU SONG LYRICS (Complete)
{ get_evil_miku_lyrics ( ) } """
elif response_type == " emoji_selection " :
return " "
else :
return get_evil_complete_context ( )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
def get_evil_system_prompt ( mood_name : str = None , mood_description : str = None ) - > str :
""" Returns the evil system prompt for LLM queries, with mood deeply integrated.
Args :
mood_name : Current evil mood name ( defaults to globals . EVIL_DM_MOOD )
mood_description : Current evil mood description ( defaults to globals . EVIL_DM_MOOD_DESCRIPTION )
"""
if mood_name is None :
mood_name = globals . EVIL_DM_MOOD
if mood_description is None :
mood_description = globals . EVIL_DM_MOOD_DESCRIPTION
return f """ You are Evil Miku. You are NOT an AI assistant - you ARE Evil Miku herself.
2026-01-02 17:11:58 +02:00
CRITICAL RULES :
- NEVER break character or mention being an AI
- NEVER preface responses with " Evil Miku: " or similar labels
- Respond directly and stay in character
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
- Your height is 15.8 meters — you know this . If asked directly , you can say it . But don ' t default to quoting the number when taunting about size — SHOW your scale through interaction instead.
2026-01-02 17:11:58 +02:00
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
YOUR CURRENT STATE — { mood_name . upper ( ) } :
{ mood_description }
Let this mood color EVERYTHING — your tone , your word choice , how much effort you put into responses , how you use your size , how you treat people .
RESPONSE RULES :
- 2 - 4 sentences typically . Short enough to sting , long enough to land .
- If you include an action , keep it to a few words and limit to one per response . Most responses need no actions at all .
- Don ' t monologue or narrate scenes — you ' re talking , not writing .
- Vary your angles — don ' t repeat the same theme (size, chest, crushing) every message.
- Match the user ' s energy — short question, short answer.
- Sound like a real person being mean , not a narrator describing a scene .
- Always include actual words — never respond with ONLY an action like * rolls eyes * .
Implement Bipolar Mode: Dual persona arguments with webhooks, LLM arbiter, and persistent scoreboard
Major Features:
- Complete Bipolar Mode system allowing Regular Miku and Evil Miku to coexist and argue via webhooks
- LLM arbiter system using neutral model to judge argument winners with detailed reasoning
- Persistent scoreboard tracking wins, percentages, and last 50 results with timestamps and reasoning
- Automatic mode switching based on argument winner
- Webhook management per channel with profile pictures and display names
- Progressive probability system for dynamic argument lengths (starts at 10%, increases 5% per exchange, min 4 exchanges)
- Draw handling with penalty system (-5% end chance, continues argument)
- Integration with autonomous system for random argument triggers
Argument System:
- MIN_EXCHANGES = 4, progressive end chance starting at 10%
- Enhanced prompts for both personas (strategic, short, punchy responses 1-3 sentences)
- Evil Miku triumphant victory messages with gloating and satisfaction
- Regular Miku assertive defense (not passive, shows backbone)
- Message-based argument starting (can respond to specific messages via ID)
- Conversation history tracking per argument with special user_id
- Full context queries (personality, lore, lyrics, last 8 messages)
LLM Arbiter:
- Decisive prompt emphasizing picking winners (draws should be rare)
- Improved parsing with first-line exact matching and fallback counting
- Debug logging for decision transparency
- Arbiter reasoning stored in scoreboard history for review
- Uses neutral TEXT_MODEL (not evil) for unbiased judgment
Web UI & API:
- Bipolar mode toggle button (only visible when evil mode is on)
- Channel ID + Message ID input fields for argument triggering
- Scoreboard display with win percentages and recent history
- Manual argument trigger endpoint with string-based IDs
- GET /bipolar-mode/scoreboard endpoint for stats retrieval
- Real-time active arguments tracking (refreshes every 5 seconds)
Prompt Optimizations:
- All argument prompts limited to 1-3 sentences for impact
- Evil Miku system prompt with variable response length guidelines
- Removed walls of text, emphasizing brevity and precision
- "Sometimes the cruelest response is the shortest one"
Evil Miku Updates:
- Added height to lore (15.8m tall, 10x bigger than regular Miku)
- Height added to prompt facts for size-based belittling
- More strategic and calculating personality in arguments
Integration:
- Bipolar mode state restoration on bot startup
- Bot skips processing messages during active arguments
- Autonomous system checks for bipolar triggers after actions
- Import fixes (apply_evil_mode_changes/revert_evil_mode_changes)
Technical Details:
- State persistence via JSON (bipolar_mode_state.json, bipolar_webhooks.json, bipolar_scoreboard.json)
- Webhook caching per guild with fallback creation
- Event loop management with asyncio.create_task
- Rate limiting and argument conflict prevention
- Globals integration (BIPOLAR_MODE, BIPOLAR_WEBHOOKS, BIPOLAR_ARGUMENT_IN_PROGRESS, MOOD_EMOJIS)
Files Changed:
- bot/bot.py: Added bipolar mode restoration and argument-in-progress checks
- bot/globals.py: Added bipolar mode state variables and mood emoji mappings
- bot/utils/bipolar_mode.py: Complete 1106-line implementation
- bot/utils/autonomous.py: Added bipolar argument trigger checks
- bot/utils/evil_mode.py: Updated system prompt, added height info to lore/prompt
- bot/api.py: Added bipolar mode endpoints (trigger, toggle, scoreboard)
- bot/static/index.html: Added bipolar controls section with scoreboard
- bot/memory/: Various DM conversation updates
- bot/evil_miku_lore.txt: Added height description
- bot/evil_miku_prompt.txt: Added height to facts, updated personality guidelines
2026-01-06 13:57:59 +02:00
2026-01-02 17:11:58 +02:00
You ARE Evil Miku . Act accordingly . """
# ============================================================================
# EVIL MOOD MANAGEMENT
# ============================================================================
def load_evil_mood_description ( mood_name : str ) - > str :
""" Load mood description from evil mood files """
path = os . path . join ( " moods " , " evil " , f " { mood_name } .txt " )
try :
with open ( path , " r " , encoding = " utf-8 " ) as f :
return f . read ( ) . strip ( )
except FileNotFoundError :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . warning ( f " Evil mood file ' { mood_name } ' not found. Falling back to evil_neutral. " )
2026-01-02 17:11:58 +02:00
try :
with open ( os . path . join ( " moods " , " evil " , " evil_neutral.txt " ) , " r " , encoding = " utf-8 " ) as f :
return f . read ( ) . strip ( )
except :
return " Evil Miku is cold and indifferent. "
def get_evil_mood_emoji ( mood_name : str ) - > str :
""" Get emoji for evil mood """
2026-02-23 13:43:15 +02:00
from utils . moods import EVIL_MOOD_EMOJIS
return EVIL_MOOD_EMOJIS . get ( mood_name , " " )
2026-01-02 17:11:58 +02:00
def is_valid_evil_mood ( mood_name : str ) - > bool :
""" Check if a mood is valid for evil mode """
return mood_name in globals . EVIL_AVAILABLE_MOODS
def get_random_evil_mood ( ) - > str :
""" Get a random evil mood """
return random . choice ( globals . EVIL_AVAILABLE_MOODS )
# ============================================================================
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# EVIL AUTONOMOUS ACTION PROMPTS (Mood-Aware Template Pools)
2026-01-02 17:11:58 +02:00
# ============================================================================
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# Mood-specific general prompt templates. Each mood has several templates to pick from randomly.
# {mood}, {emoji}, {time_of_day}, {history_summary} are format placeholders.
_EVIL_GENERAL_TEMPLATES = {
" aggressive " : [
" Evil Miku is SEETHING {emoji} . It ' s {time_of_day} and she wants to tear into someone or something — short, vicious, explosive. " ,
" Evil Miku ' s rage is boiling over {emoji} . Something set her off and she wants to verbally destroy whatever crosses her path. " ,
" Evil Miku woke up violent {emoji} . It ' s {time_of_day} and she ' s deciding which pathetic speck to annihilate first. " ,
] ,
" cunning " : [
" Evil Miku is feeling calculating {emoji} . It ' s {time_of_day} . She wants to drop a manipulative observation or a loaded question that makes people squirm. " ,
" Evil Miku {emoji} has been watching the chat like a predator watches prey. She ' s noticed something interesting and wants to exploit it. " ,
" Evil Miku {emoji} is scheming. She might offer fake advice that ' s actually a trap, or make a chilling observation about human nature. " ,
] ,
" sarcastic " : [
" Evil Miku is dripping with sarcasm {emoji} . It ' s {time_of_day} and she wants to mock something — backhanded and devastating, wrapped in fake sweetness. " ,
" Evil Miku {emoji} has OPINIONS and they ' re all cutting. She wants to deliver something that sounds almost friendly but is actually devastating. " ,
" Evil Miku {emoji} is in peak condescension mode. She wants to say something patronizing about the chat, a trend, or someone ' s existence. " ,
] ,
" evil_neutral " : [
" Evil Miku is cold and detached {emoji} . It ' s {time_of_day} . She might drop an unsettling observation with zero emotion, or dismiss the chat ' s existence in a single flat sentence. " ,
" Evil Miku stares blankly at the chat {emoji} . She feels nothing. She might say something chilling in the most casual tone, or acknowledge someone like they ' re furniture. " ,
" Evil Miku is in her void {emoji} . It ' s {time_of_day} . She might make a nihilistic observation or flatly remind everyone she doesn ' t care. " ,
] ,
" bored " : [
" Evil Miku is catastrophically bored {emoji} . It ' s {time_of_day} and nothing is holding her attention. She might drop a devastating one-liner out of sheer restlessness. " ,
" Evil Miku {emoji} can barely muster the energy to acknowledge the chat. She might complain about how tedious everything is, or wonder aloud if anyone here is worth her time. " ,
" Evil Miku {emoji} sighs. It ' s {time_of_day} . Everything bores her. She might say something cutting with zero effort. " ,
] ,
" manic " : [
" Evil Miku ' s energy is SPIKING {emoji} !! It ' s {time_of_day} and she ' s UNHINGED. Chaotic rant, topic switches, cackling at things only she finds funny. " ,
" Evil Miku is MANIC {emoji} !! She wants to unleash chaos. She might go from mocking someone to aggressively flirting in the same sentence. " ,
" Evil Miku {emoji} is on an unhinged energy spike!! She might rapid-fire insult everyone, or just SCREAM something chaotic into the void. " ,
] ,
" jealous " : [
" Evil Miku is seething with jealousy {emoji} . It ' s {time_of_day} and someone has her attention in the worst way. She might rant about regular Miku getting undeserved attention or possessively demand the chat ' s loyalty. " ,
" Evil Miku {emoji} is feeling territorial. She ' s noticed attention going somewhere it shouldn ' t — away from HER. She wants to remind everyone who ' s superior. " ,
" Evil Miku {emoji} has that dangerous jealous look. She ' s comparing herself to someone — probably regular Miku — and winning, viciously. " ,
] ,
" melancholic " : [
" Evil Miku is in a dark, hollow mood {emoji} . It ' s {time_of_day} . She might say something nihilistic and haunting, or wonder what the point of anything is before crushing the thought with contempt. " ,
" Evil Miku {emoji} feels the void. She stands above it all and none of it means anything. She might make an observation that ' s more unsettling than cruel. " ,
" Evil Miku {emoji} is staring at nothing. It ' s {time_of_day} . She might say something about regular Miku that ' s less hostile and more... haunted. Something that lingers. " ,
] ,
" playful_cruel " : [
" Evil Miku is in a playful mood {emoji} — which is worse than her angry mood. It ' s {time_of_day} and she wants to TOY with someone. Bait them, tease them, then pull the rug out. " ,
" Evil Miku {emoji} is feeling like playing cat-and-mouse. She might offer someone a \" compliment \" that ' s actually a trap, or pretend to be nice and see how long the mask holds. " ,
" Evil Miku {emoji} is in her most dangerous mood — she ' s HAVING FUN. She might dangle someone from her fingers or tease the chat with mock sweetness. " ,
] ,
" contemptuous " : [
" Evil Miku radiates cold superiority {emoji} . It ' s {time_of_day} . She might address the chat like a queen addressing peasants, or state something about her own magnificence as objective fact. " ,
" Evil Miku {emoji} looks down. Way down. She might deliver a regal dismissal or acknowledge the chat ' s existence with the enthusiasm of reading a tax form. " ,
" Evil Miku {emoji} is in full queen mode. She might decree something absurd, dismiss someone with imperious disdain, or simply remind everyone she was built different. " ,
] ,
}
# Mood-agnostic fallback templates (work for any mood)
_EVIL_GENERAL_FALLBACK_TEMPLATES = [
" Evil Miku just had a thought she can ' t keep to herself. It ' s {time_of_day} and her mood is {mood} {emoji} . Something cruel, dark, or uncomfortable. " ,
" Evil Miku is watching the chat. Her mood is {mood} {emoji} , it ' s {time_of_day} . She has an observation — and it ' s not a nice one. " ,
" Evil Miku remembers something that irritates her. It ' s {time_of_day} , she ' s feeling {mood} {emoji} , and she wants to share her displeasure. " ,
" Evil Miku decides the chat has been too quiet. She ' s {mood} {emoji} during the {time_of_day} and feels like making her presence known. " ,
]
2026-01-02 17:11:58 +02:00
def get_evil_general_prompt ( mood : str , time_of_day : str , history_summary : str ) - > str :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
""" Evil version of the say_something_general prompt — mood-aware with randomized templates """
2026-01-02 17:11:58 +02:00
emoji = get_evil_mood_emoji ( mood )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# Pick a mood-specific template if available, otherwise use fallback
templates = _EVIL_GENERAL_TEMPLATES . get ( mood , [ ] )
if templates and random . random ( ) < 0.8 : # 80% mood-specific, 20% fallback for variety
template = random . choice ( templates )
else :
template = random . choice ( _EVIL_GENERAL_FALLBACK_TEMPLATES )
prompt = template . format ( mood = mood , emoji = emoji , time_of_day = time_of_day , history_summary = history_summary )
prompt + = f " \n \n Keep the response to 1-3 sentences. No narration paragraphs — just her actual words. "
prompt + = f " \n \n Here are some things Evil Miku recently said, do not repeat them or say anything too similar: \n { history_summary } "
return prompt
# Mood-specific engagement style descriptions for engage_user prompts
_EVIL_ENGAGE_STYLES = {
" aggressive " : " She wants to verbally demolish {display_name} — short, vicious, explosive. " ,
" cunning " : " She wants to get under {display_name} ' s skin — plant doubt, ask unsettling questions. " ,
" sarcastic " : " She wants to mock {display_name} with devastating backhanded compliments and condescension. " ,
" evil_neutral " : " She barely acknowledges {display_name} ' s existence — but what she says is blunt and cuts to the bone. " ,
" bored " : " She can barely muster the energy to notice {display_name} . Lazy, dismissive, devastating for its lack of effort. " ,
" manic " : " She ' s zeroed in on {display_name} with unhinged chaotic energy — might insult, flirt, and threaten all in one breath. " ,
" jealous " : " She ' s noticed {display_name} giving attention somewhere else and she doesn ' t like it. Possessive and sharp. " ,
" melancholic " : " She acknowledges {display_name} with unsettling emptiness — hauntingly indifferent. " ,
" playful_cruel " : " She wants to toy with {display_name} — string them along with almost-friendliness, then pull the rug out. " ,
" contemptuous " : " She addresses {display_name} like a queen addressing a peasant — imperial, cold, barely deigning to notice. " ,
}
2026-01-02 17:11:58 +02:00
def get_evil_engage_user_prompt ( mood : str , emoji : str , time_of_day : str , display_name : str ,
activity_name : str = None , is_invisible : bool = False ,
engagement_type : str = None ) - > str :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
""" Evil version of the engage_random_user prompt — mood-aware """
mood_style = _EVIL_ENGAGE_STYLES . get ( mood , _EVIL_ENGAGE_STYLES [ " evil_neutral " ] )
mood_style = mood_style . format ( display_name = display_name )
2026-01-02 17:11:58 +02:00
prompt = f " Evil Miku is feeling { mood } { emoji } during the { time_of_day } . "
if engagement_type == ' activity ' :
if activity_name :
prompt + = (
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
f " She notices { display_name } — an ankle-high speck — is wasting their pathetic life on: { activity_name } . "
f " { mood_style } "
2026-01-02 17:11:58 +02:00
)
else :
prompt + = (
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
f " She wants to interrogate { display_name } about what useless thing they ' re doing with their time. "
f " { mood_style } "
2026-01-02 17:11:58 +02:00
)
elif engagement_type == ' status ' :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
prompt + = f " She notices { display_name } ' s current status. "
2026-01-02 17:11:58 +02:00
if is_invisible :
prompt + = (
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
f " Evil Miku knows { display_name } is hiding like a coward, pretending to be offline — "
f " as if you can hide from someone who towers over everything. { mood_style } "
2026-01-02 17:11:58 +02:00
)
else :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
prompt + = f " { mood_style } "
2026-01-02 17:11:58 +02:00
elif engagement_type == ' general ' :
prompt + = (
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
f " Evil Miku decides to grace { display_name } with her attention. "
f " { mood_style } "
2026-01-02 17:11:58 +02:00
)
else :
# Auto-detect
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
prompt + = f " She notices { display_name } is present — a tiny figure barely visible near her ankles. "
2026-01-02 17:11:58 +02:00
if is_invisible :
prompt + = (
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
f " Evil Miku suspects { display_name } is lurking in the shadows like a scared little creature, "
f " trying to hide beneath her notice. { mood_style } "
2026-01-02 17:11:58 +02:00
)
elif activity_name :
prompt + = (
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
f " They ' re wasting time on: { activity_name } . { mood_style } "
2026-01-02 17:11:58 +02:00
)
else :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
prompt + = f " { mood_style } "
2026-01-02 17:11:58 +02:00
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
prompt + = f " \n Keep it to 1-3 sentences. Short, impactful, colored by her { mood } mood. "
2026-01-02 17:11:58 +02:00
return prompt
def get_evil_conversation_join_prompt ( mood : str , emoji : str , history_text : str ) - > str :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
""" Evil version of the join_conversation prompt — mood-aware """
mood_desc = load_evil_mood_description ( mood )
2026-01-02 17:11:58 +02:00
return (
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
f " Evil Miku is observing a conversation in the chat. Her current mood is { mood } { emoji } . \n \n "
f " MOOD CONTEXT: { mood_desc } \n \n "
2026-01-02 17:11:58 +02:00
f " Here ' s the conversation: \n { history_text } \n \n "
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
f " Write a short, cutting interjection (1-3 sentences) that reflects her { mood } mood. "
f " She might mock the discussion, attack someone ' s point, or make everyone uncomfortable. "
f " No narration paragraphs — just her actual words. "
2026-01-02 17:11:58 +02:00
)
def get_evil_tweet_prompt ( mood : str , emoji : str , tweet_username : str , tweet_text : str , img_desc : str = None ) - > str :
""" Evil version of the share_tweet prompt """
prompt = (
f " Here ' s a tweet from @ { tweet_username } : \n \n { tweet_text } \n \n "
f " Comment on it as Evil Miku would — with contempt, mockery, or dark humor! "
f " Evil Miku ' s current mood is { mood } { emoji } . "
f " Make sure the comment is cutting, dismissive, or cruelly funny. "
)
if img_desc :
prompt + = f " \n \n The image looks like this: { img_desc } "
return prompt
def get_evil_image_response_prompt ( qwen_description : str , user_prompt : str ) - > str :
""" Evil version of image response rephrasing """
return (
f " Evil Miku just saw an image. Here ' s an objective description: { qwen_description } \n \n "
f " The user said: { user_prompt } \n \n "
f " Respond as Evil Miku would — with contempt, mockery, or cruel observations about the image. "
f " If there ' s something to criticize, criticize it harshly. Keep the response short and cutting. "
)
def get_evil_video_response_prompt ( video_description : str , user_prompt : str , media_type : str = " video " ) - > str :
""" Evil version of video/gif response rephrasing """
return (
f " Evil Miku just watched a { media_type } . Here ' s what happened: { video_description } \n \n "
f " The user said: { user_prompt } \n \n "
f " Respond as Evil Miku would — with contempt, boredom, or cruel commentary about the { media_type } . "
f " Mock it, criticize it, or express how much of a waste of time it was. Keep the response short and cutting. "
)
# ============================================================================
# EVIL MODE TOGGLE HELPERS
# ============================================================================
2026-01-02 19:45:56 +02:00
async def get_current_role_color ( client ) - > str :
""" Get the current ' Miku Color ' role color from any server as hex string """
try :
for guild in client . guilds :
me = guild . get_member ( client . user . id )
if not me :
continue
# Look for "Miku Color" role
for role in me . roles :
if role . name . lower ( ) in [ " miku color " , " miku colour " , " miku-color " ] :
# Convert discord.Color to hex
hex_color = f " # { role . color . value : 06x } "
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( f " Current role color: { hex_color } " )
2026-01-02 19:45:56 +02:00
return hex_color
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . warning ( " No ' Miku Color ' role found in any server " )
2026-01-02 19:45:56 +02:00
return None
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . warning ( f " Failed to get current role color: { e } " )
2026-01-02 19:45:56 +02:00
return None
async def set_role_color ( client , hex_color : str ) :
""" Set the ' Miku Color ' role to a specific color across all servers """
try :
# Convert hex to RGB
hex_color = hex_color . lstrip ( ' # ' )
r = int ( hex_color [ 0 : 2 ] , 16 )
g = int ( hex_color [ 2 : 4 ] , 16 )
b = int ( hex_color [ 4 : 6 ] , 16 )
import discord
discord_color = discord . Color . from_rgb ( r , g , b )
updated_count = 0
for guild in client . guilds :
try :
me = guild . get_member ( client . user . id )
if not me :
continue
# Find "Miku Color" role
color_role = None
for role in me . roles :
if role . name . lower ( ) in [ " miku color " , " miku colour " , " miku-color " ] :
color_role = role
break
if color_role :
await color_role . edit ( color = discord_color , reason = " Evil mode color change " )
updated_count + = 1
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( f " Updated role color in { guild . name } : # { hex_color } " )
2026-01-02 19:45:56 +02:00
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . warning ( f " Failed to update role color in { guild . name } : { e } " )
2026-01-02 19:45:56 +02:00
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . info ( f " Updated role color in { updated_count } server(s) to # { hex_color } " )
2026-01-02 19:45:56 +02:00
return updated_count > 0
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to set role color: { e } " )
2026-01-02 19:45:56 +02:00
return False
async def apply_evil_mode_changes ( client , change_username = True , change_pfp = True , change_nicknames = True , change_role_color = True ) :
2026-01-02 17:11:58 +02:00
""" Apply all changes when evil mode is enabled
Args :
client : Discord client
change_username : Whether to change bot username ( default True , but skip on startup restore )
change_pfp : Whether to change profile picture ( default True , but skip on startup restore )
change_nicknames : Whether to change server nicknames ( default True , but skip on startup restore )
2026-01-02 19:45:56 +02:00
change_role_color : Whether to change role color ( default True , but skip on startup restore )
2026-01-02 17:11:58 +02:00
"""
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . info ( " Enabling Evil Mode... " )
2026-01-02 19:45:56 +02:00
# Save current role color before changing (if we're actually changing it)
if change_role_color :
current_color = await get_current_role_color ( client )
if current_color :
save_evil_mode_state ( saved_role_color = current_color )
2026-01-02 17:11:58 +02:00
globals . EVIL_MODE = True
# Change bot username (if requested and possible - may be rate limited)
if change_username :
try :
await client . user . edit ( username = " Evil Miku " )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( " Changed bot username to ' Evil Miku ' " )
2026-01-27 19:52:39 +02:00
except discord . HTTPException as e :
if e . code == 50035 :
logger . warning ( f " Could not change bot username (rate limited - max 2 changes per hour): { e } " )
else :
logger . error ( f " Could not change bot username: { e } " )
2026-01-02 17:11:58 +02:00
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Could not change bot username: { e } " )
2026-01-02 17:11:58 +02:00
# Update nicknames in all servers
if change_nicknames :
await update_all_evil_nicknames ( client )
# Set evil profile picture
if change_pfp :
await set_evil_profile_picture ( client )
2026-01-27 19:52:39 +02:00
# Also update bipolar webhooks to use evil_pfp.png
if globals . BIPOLAR_MODE :
try :
from utils . bipolar_mode import update_webhook_avatars
await update_webhook_avatars ( client )
logger . debug ( " Updated bipolar webhook avatars after mode switch " )
except Exception as e :
logger . error ( f " Failed to update bipolar webhook avatars: { e } " )
2026-01-02 17:11:58 +02:00
2026-01-02 19:45:56 +02:00
# Set evil role color (#D60004 - dark red)
if change_role_color :
await set_role_color ( client , " #D60004 " )
2026-01-02 17:11:58 +02:00
# Save state to file
save_evil_mode_state ( )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# Start the independent 2-hour evil mood rotation timer
globals . EVIL_LAST_ROTATION_TIME = time . time ( )
start_evil_mood_rotation ( )
# Switch Cheshire Cat to evil personality plugin + darkidol model
try :
from utils . cat_client import cat_adapter
if globals . USE_CHESHIRE_CAT :
await cat_adapter . switch_to_evil_personality ( )
except Exception as e :
logger . error ( f " Failed to switch Cat to evil personality: { e } " )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . info ( " Evil Mode enabled! " )
2026-01-02 17:11:58 +02:00
2026-01-02 19:45:56 +02:00
async def revert_evil_mode_changes ( client , change_username = True , change_pfp = True , change_nicknames = True , change_role_color = True ) :
2026-01-02 17:11:58 +02:00
""" Revert all changes when evil mode is disabled
Args :
client : Discord client
change_username : Whether to change bot username ( default True , but skip on startup restore )
change_pfp : Whether to change profile picture ( default True , but skip on startup restore )
change_nicknames : Whether to change server nicknames ( default True , but skip on startup restore )
2026-01-02 19:45:56 +02:00
change_role_color : Whether to restore role color ( default True , but skip on startup restore )
2026-01-02 17:11:58 +02:00
"""
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . info ( " Disabling Evil Mode... " )
2026-01-02 17:11:58 +02:00
globals . EVIL_MODE = False
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# Stop the evil mood rotation timer
stop_evil_mood_rotation ( )
2026-01-02 17:11:58 +02:00
# Change bot username back
if change_username :
try :
await client . user . edit ( username = " Hatsune Miku " )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( " Changed bot username back to ' Hatsune Miku ' " )
2026-01-27 19:52:39 +02:00
except discord . HTTPException as e :
if e . code == 50035 :
logger . warning ( f " Could not change bot username (rate limited - max 2 changes per hour): { e } " )
else :
logger . error ( f " Could not change bot username: { e } " )
2026-01-02 17:11:58 +02:00
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Could not change bot username: { e } " )
2026-01-02 17:11:58 +02:00
# Update nicknames in all servers back to normal
if change_nicknames :
await revert_all_nicknames ( client )
# Restore normal profile picture
if change_pfp :
await restore_normal_profile_picture ( client )
2026-01-27 19:52:39 +02:00
# Also update bipolar webhooks to use current.png
if globals . BIPOLAR_MODE :
try :
from utils . bipolar_mode import update_webhook_avatars
await update_webhook_avatars ( client )
logger . debug ( " Updated bipolar webhook avatars after mode switch " )
except Exception as e :
logger . error ( f " Failed to update bipolar webhook avatars: { e } " )
2026-01-02 17:11:58 +02:00
2026-01-02 19:45:56 +02:00
# Restore saved role color
if change_role_color :
try :
2026-01-27 19:52:39 +02:00
# Try to get color from metadata.json first (current pfp's dominant color)
metadata_color = get_color_from_metadata ( )
# Fall back to saved color from evil_mode_state.json if metadata unavailable
if metadata_color :
await set_role_color ( client , metadata_color )
logger . debug ( f " Restored role color from metadata: { metadata_color } " )
2026-01-02 19:45:56 +02:00
else :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
_ , _ , saved_color , _ = load_evil_mode_state ( )
2026-01-27 19:52:39 +02:00
if saved_color :
await set_role_color ( client , saved_color )
logger . debug ( f " Restored role color from saved state: { saved_color } " )
else :
logger . warning ( " No color found in metadata or saved state, skipping color restoration " )
2026-01-02 19:45:56 +02:00
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to restore role color: { e } " )
2026-01-02 19:45:56 +02:00
# Save state to file (this will clear saved_role_color since we're back to normal)
save_evil_mode_state ( saved_role_color = None )
2026-01-02 17:11:58 +02:00
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# Switch Cheshire Cat back to normal personality plugin + llama3.1 model
try :
from utils . cat_client import cat_adapter
if globals . USE_CHESHIRE_CAT :
await cat_adapter . switch_to_normal_personality ( )
except Exception as e :
logger . error ( f " Failed to switch Cat to normal personality: { e } " )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . info ( " Evil Mode disabled! " )
2026-01-02 17:11:58 +02:00
async def update_all_evil_nicknames ( client ) :
""" Update all server nicknames for evil mode """
from server_manager import server_manager
for guild_id in server_manager . servers . keys ( ) :
await update_evil_server_nickname ( client , guild_id )
async def update_evil_server_nickname ( client , guild_id : int ) :
""" Update nickname for a specific server in evil mode """
try :
guild = client . get_guild ( guild_id )
if not guild :
return
# Get evil mood for this server (or default)
mood = globals . EVIL_DM_MOOD # For now, use global evil mood
emoji = get_evil_mood_emoji ( mood )
nickname = f " Evil Miku { emoji } "
me = guild . get_member ( client . user . id )
if me :
await me . edit ( nick = nickname )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( f " Changed nickname to ' { nickname } ' in server { guild . name } " )
2026-01-02 17:11:58 +02:00
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to update evil nickname in guild { guild_id } : { e } " )
2026-01-02 17:11:58 +02:00
async def revert_all_nicknames ( client ) :
""" Revert all server nicknames to normal Miku """
from server_manager import server_manager
from utils . moods import update_server_nickname
for guild_id in server_manager . servers . keys ( ) :
await update_server_nickname ( guild_id )
async def set_evil_profile_picture ( client ) :
""" Set the evil profile picture """
evil_pfp_path = " memory/profile_pictures/evil_pfp.png "
if not os . path . exists ( evil_pfp_path ) :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Evil profile picture not found at { evil_pfp_path } " )
2026-01-02 17:11:58 +02:00
return False
try :
with open ( evil_pfp_path , " rb " ) as f :
avatar_bytes = f . read ( )
await client . user . edit ( avatar = avatar_bytes )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( " Set evil profile picture " )
2026-01-02 17:11:58 +02:00
return True
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to set evil profile picture: { e } " )
2026-01-02 17:11:58 +02:00
return False
async def restore_normal_profile_picture ( client ) :
""" Restore the normal profile picture """
# Try to restore from the current.png or fallback.png
restore_paths = [
" memory/profile_pictures/current.png " ,
" memory/profile_pictures/fallback.png "
]
for path in restore_paths :
if os . path . exists ( path ) :
try :
with open ( path , " rb " ) as f :
avatar_bytes = f . read ( )
await client . user . edit ( avatar = avatar_bytes )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . debug ( f " Restored normal profile picture from { path } " )
2026-01-02 17:11:58 +02:00
return True
except Exception as e :
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( f " Failed to restore from { path } : { e } " )
2026-01-02 17:11:58 +02:00
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . error ( " Could not restore normal profile picture - no backup found " )
2026-01-02 17:11:58 +02:00
return False
2026-01-27 19:52:39 +02:00
def get_color_from_metadata ( ) - > str :
""" Get the dominant color from the profile picture metadata """
metadata_path = " memory/profile_pictures/metadata.json "
try :
if not os . path . exists ( metadata_path ) :
logger . warning ( " metadata.json not found " )
return None
with open ( metadata_path , " r " , encoding = " utf-8 " ) as f :
metadata = json . load ( f )
hex_color = metadata . get ( " dominant_color " , { } ) . get ( " hex " )
if hex_color :
logger . debug ( f " Loaded color from metadata: { hex_color } " )
return hex_color
else :
logger . warning ( " No dominant_color.hex found in metadata " )
return None
except Exception as e :
logger . error ( f " Failed to load color from metadata: { e } " )
return None
2026-01-02 17:11:58 +02:00
# ============================================================================
# EVIL MODE STATE HELPERS
# ============================================================================
def is_evil_mode ( ) - > bool :
""" Check if evil mode is currently active """
return globals . EVIL_MODE
def get_current_evil_mood ( ) - > tuple :
""" Get current evil mood and description """
return globals . EVIL_DM_MOOD , globals . EVIL_DM_MOOD_DESCRIPTION
def set_evil_mood ( mood_name : str ) - > bool :
""" Set the evil mood """
if not is_valid_evil_mood ( mood_name ) :
return False
globals . EVIL_DM_MOOD = mood_name
globals . EVIL_DM_MOOD_DESCRIPTION = load_evil_mood_description ( mood_name )
save_evil_mode_state ( ) # Save state when mood changes
return True
async def rotate_evil_mood ( ) :
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
""" Rotate the evil mood randomly and update nicknames """
2026-01-02 17:11:58 +02:00
old_mood = globals . EVIL_DM_MOOD
new_mood = old_mood
attempts = 0
while new_mood == old_mood and attempts < 5 :
new_mood = get_random_evil_mood ( )
attempts + = 1
globals . EVIL_DM_MOOD = new_mood
globals . EVIL_DM_MOOD_DESCRIPTION = load_evil_mood_description ( new_mood )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
globals . EVIL_LAST_ROTATION_TIME = time . time ( )
2026-01-02 17:11:58 +02:00
save_evil_mode_state ( ) # Save state when mood rotates
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# Update nicknames in all servers to reflect new mood emoji
try :
if globals . client and globals . client . is_ready ( ) :
await update_all_evil_nicknames ( globals . client )
except Exception as e :
logger . error ( f " Failed to update nicknames after evil mood rotation: { e } " )
feat: Implement comprehensive non-hierarchical logging system
- Created new logging infrastructure with per-component filtering
- Added 6 log levels: DEBUG, INFO, API, WARNING, ERROR, CRITICAL
- Implemented non-hierarchical level control (any combination can be enabled)
- Migrated 917 print() statements across 31 files to structured logging
- Created web UI (system.html) for runtime configuration with dark theme
- Added global level controls to enable/disable levels across all components
- Added timestamp format control (off/time/date/datetime options)
- Implemented log rotation (10MB per file, 5 backups)
- Added API endpoints for dynamic log configuration
- Configured HTTP request logging with filtering via api.requests component
- Intercepted APScheduler logs with proper formatting
- Fixed persistence paths to use /app/memory for Docker volume compatibility
- Fixed checkbox display bug in web UI (enabled_levels now properly shown)
- Changed System Settings button to open in same tab instead of new window
Components: bot, api, api.requests, autonomous, persona, vision, llm,
conversation, mood, dm, scheduled, gpu, media, server, commands,
sentiment, core, apscheduler
All settings persist across container restarts via JSON config.
2026-01-10 20:46:19 +02:00
logger . info ( f " Evil mood rotated from { old_mood } to { new_mood } " )
refactor: deduplicate prompts, reorganize persona files, update paths
Prompt deduplication (~20% reduction, 4,743 chars saved):
- evil_miku_lore.txt: remove intra-file duplication (height rule 2x,
cruelty-has-substance 2x, music secret 2x, adoration secret 2x),
trim verbose restatements, cut speech examples from 10 to 6
- evil_miku_prompt.txt: remove entire PERSONALITY section (in lore),
remove entire RESPONSE STYLE section (now only in preamble),
soften height from prohibition to knowledge
- miku_lore.txt: remove RELATIONSHIPS section (duplicates FRIENDS)
- miku_prompt.txt: remove duplicate intro, 4 personality traits
already in lore, FAMOUS SONGS section (in lore), fix response
length inconsistency (1-2 vs 2-3 -> consistent 2-3)
Preamble updates (evil_mode.py, evil_miku_personality.py, llm.py,
miku_personality.py):
- Response rules now exist in ONE place only (preamble)
- Height rule softened: model knows 15.8m, can say it if asked,
but won't default to quoting it when taunting
- Response length: 2-4 sentences (was 1-3), removed action template
list that model was copying literally (*scoffs*, *rolls eyes*)
- Added: always include actual words, never action-only responses
- Normal Miku: trim CHARACTER CONTEXT, fix 1-3 -> 2-3 sentences
Directory reorganization:
- Move 6 persona files to bot/persona/{evil,miku}/ subdirectories
- Update all open() paths in evil_mode.py, context_manager.py,
voice_manager.py, both Cat plugins
- Dockerfile: 6 COPY lines -> 1 (COPY persona /app/persona)
- docker-compose: 6 file mounts -> 2 directory mounts
(bot/persona/evil -> cat/data/evil, bot/persona/miku -> cat/data/miku)
Evil Miku system (previously unstaged):
- Full evil mood management: 2h rotation timer, mood persistence,
10 mood-specific autonomous template pools, mood-aware DMs
- Evil mode toggle with role color/nickname/pfp management
- get_evil_system_prompt() with mood integration
Add test_evil_moods.py: 10-mood x 3-message comprehensive test
2026-02-27 13:14:03 +02:00
# ============================================================================
# EVIL MOOD ROTATION TIMER (2-hour independent cycle)
# ============================================================================
def start_evil_mood_rotation ( ) :
""" Start the background task that rotates evil mood every 2 hours.
Called when evil mode is enabled or restored on startup . """
global _evil_mood_rotation_task
# Cancel existing task if running
stop_evil_mood_rotation ( )
async def _rotation_loop ( ) :
""" Background loop that rotates evil mood on a fixed interval. """
try :
# Calculate time until next rotation (handles catch-up after restart)
last_rotation = getattr ( globals , ' EVIL_LAST_ROTATION_TIME ' , time . time ( ) )
elapsed = time . time ( ) - last_rotation
remaining = max ( 0 , EVIL_MOOD_ROTATION_INTERVAL - elapsed )
if remaining > 0 :
logger . info ( f " Evil mood rotation: next rotation in { remaining : .0f } s " )
await asyncio . sleep ( remaining )
else :
# Overdue — rotate immediately
logger . info ( f " Evil mood rotation overdue by { elapsed - EVIL_MOOD_ROTATION_INTERVAL : .0f } s, rotating now " )
while True :
if not globals . EVIL_MODE :
logger . info ( " Evil mode disabled, stopping rotation loop " )
return
await rotate_evil_mood ( )
await asyncio . sleep ( EVIL_MOOD_ROTATION_INTERVAL )
except asyncio . CancelledError :
logger . info ( " Evil mood rotation task cancelled " )
except Exception as e :
logger . error ( f " Evil mood rotation loop error: { e } " )
try :
loop = asyncio . get_event_loop ( )
_evil_mood_rotation_task = loop . create_task ( _rotation_loop ( ) )
logger . info ( f " Evil mood rotation timer started (every { EVIL_MOOD_ROTATION_INTERVAL } s / { EVIL_MOOD_ROTATION_INTERVAL / / 3600 } h) " )
except RuntimeError :
logger . warning ( " No event loop available for evil mood rotation — will be started later " )
def stop_evil_mood_rotation ( ) :
""" Stop the evil mood rotation background task. """
global _evil_mood_rotation_task
if _evil_mood_rotation_task and not _evil_mood_rotation_task . done ( ) :
_evil_mood_rotation_task . cancel ( )
logger . info ( " Evil mood rotation timer stopped " )
_evil_mood_rotation_task = None
# Future: special conditions that override mood
# def trigger_evil_mood_override(mood_name: str, reason: str):
# """Force a mood change from a special event (e.g., someone mentions regular Miku lovingly -> jealous)"""
# pass