AJ Isaacs c63913cf14 fix: anonymize usernames before LLM analysis to prevent name-based scoring bias ...

Display names like "Calm your tits" were causing the LLM to inflate toxicity
scores on completely benign messages. Usernames are now replaced with User1,
User2, etc. before sending to the LLM, then mapped back to real names in the
results.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-02-25 22:20:53 -05:00

cogs

fix: anonymize usernames before LLM analysis to prevent name-based scoring bias

2026-02-25 22:20:53 -05:00

prompts

fix: anonymize usernames before LLM analysis to prevent name-based scoring bias

2026-02-25 22:20:53 -05:00

scripts

Add announce script for posting to Discord channels

2026-02-23 10:11:27 -05:00

utils

fix: guard against malformed LLM findings in conversation validation

2026-02-25 21:38:02 -05:00

.env.example

Support hybrid LLM: local Qwen triage + OpenAI escalation

2026-02-23 12:20:07 -05:00

.gitignore

Initial commit: Breehavior Monitor Discord bot

2026-02-20 22:39:40 -05:00

bot.py

Add Wordle commentary: bot reacts to Wordle results with mode-appropriate comments

2026-02-24 17:56:05 -05:00

CLAUDE.md

refactor: extract sentiment cog into package with shared _process_finding

2026-02-25 17:06:27 -05:00

config.yaml

fix: persist last_offense_time and reset offenses after 24h

2026-02-25 11:24:38 -05:00

docker-compose.yml

Initial commit: Breehavior Monitor Discord bot

2026-02-20 22:39:40 -05:00

Dockerfile

Fix missing libgssapi-krb5-2 dependency in Docker image

2026-02-20 22:48:15 -05:00

requirements.txt

Initial commit: Breehavior Monitor Discord bot

2026-02-20 22:39:40 -05:00

Description

No description provided

1.2 MiB

Languages

Python 98.8%

Shell 0.8%

Dockerfile 0.4%