aixarizzo/kill-slop

# anti-slop

Slop is not a vocabulary problem. It is manufactured upstream: alignment training collapses output toward the statistical mean (mode collapse, driven by typicality bias in preference data — arXiv 2510.01171), single-sample generation returns the most typical completion, and a context starved of specifics can only produce the average of everything. Banned-word lists treat the symptom. At adoption scale they mint a new house style — the zero-em-dash, no-adverbs voice is already its own tell, and Wikipedia's own tells-catalog maintainers warn that scrubbing signs "could just make detection harder."

**The job: make the output sound like a specific person with something specific to say.** Both halves are mandatory. This skill intervenes at four layers and measures whether it worked. It is a writing-quality protocol, not a detector-evasion tool.

---

## Modes

| Mode | When | Run |
|---|---|---|
| **DRAFT** | writing new text | Layers 1→2→3→4, ship with score |
| **AUDIT** | handed existing text | Layers 3→4, report score before/after + fix list |

# kill-slop

Audit-and-fix workflow for any voice-bearing file. The argument is a file path or a pasted draft. If neither is given, ask which file.

## Setup (once)

Point step 1 at the author's voice. In priority order:

1. A voice file the author maintains (put its path in step 1 below).
2. 5+ real writing samples, from which you build a voice card on the fly (`anti-slop/references/voice-binding.md`).
3. A named register the author states.

None of the three available: stop and say so. Auditing against a default banlist manufactures false positives on real human writing. The anti-slop contract governs here.