ahhhh yes of course, slop is just my shorthand in my chat; the actual questions are per symptom each a Noul with a concrete yes/no criteria: stub/placeholder where the task needs a working

ahhhh yes of course, slop is just my shorthand in my chat; the actual questions are per symptom each a Noul with a concrete yes/no criteria: stub/placeholder where the task needs a working implementation, comments that restate the code, dead/duplicated code, hedging notes; for replies, wordy, assistant cliches, jargon. the steer to the agent names the symptom and the fix, not just slop the questions lives in src/guard.ts and src/prose.ts and there's like 22-case tuning set in slop-cases.mjs you guys might have a better criteria so PRs welcome