Open-Source Skill Strips AI Slop: 41 Patterns Caught Across 4 Review Gates
Roger_M_Taylor · x · 2026-09-16
Someone released a skill called anti-ai-slop that removes AI-flavored writing from Codex/Claude Code output. It catches 41 slop patterns: roughly 120 AI words and 100 phrases, plus em dashes, emoji bullets, title-case headings, watermarks, negative parallelism, colon reveals and throat-clearing openers.
It runs four gates: 1) sanitize invisible characters; 2) strip common AI words and characters; 3) an agent reviews original vs rewrite and loops back on failed reviews; 4) optional — a parallel agent on a different model generates extra review and a score.
The skill claims to work with any AI agent, and mentions using LiteLLM to wire up multiple reviewer models.
More from coding & agent
- When an LLM plans an executable agent DAG, where do you draw the trust boundary? — Repulsive_Sugar_5252 · 2026-09-16
- "Never again": user rants Nemotron 3 Ultra can't find .github and stalls for 90 minutes — dwerps · 2026-09-16
- Personal AI's real test: does your judgment survive after you close the window? — sujingshen · 2026-09-16
- Don't applaud cross-session memory yet: compaction drops decision boundaries — sujingshen · 2026-09-16
- Claude Opus 5 drew every frame of this animation using JavaScript — Many_Wave3597 · 2026-09-16
- Coding agent so fears AWS costs it wrote a test banning NAT gateways — mimi10v3 · 2026-09-16